What Works? Collaborative Police and Health interventions for mental health distress

Types of Interventions and What Works

Many collaborative interventions have been developed and piloted across the world (Macaskill et al., 2011; Martínez, 2010; Shapiro et al., 2015). Each of the inter-agency approaches tend to fall into one of three categories (Kane et al., 2017; Shapiro et al., 2015; Wood and Watson, 2017):

• Increased training on mental health distress for police officers alongside improved interaction with mental health staff
• Mental health staff working with police services to triage and offer advice and support over the phone, or face-to-face
• A co-responding police and mental health team response or mobile unit

This section goes through each of these types of intervention in turn. It discusses some examples of where these have been used, considers evaluations and provides a summary of 'what works' for each type of approach.

Increased Police Training and Liaison

Increased police training to deal with mental health distress is a key aspect of most interventions. The use of increased training is often coupled with increased and more appropriate liaison with mental health services and is apparent in interventions such as:

• Crisis Intervention Teams ( CIT)
• Mental Health Intervention Teams in Australia
• Connect Project in North Yorkshire

Studies suggest that police training tends to reduce the time police spend at the scene for a mental health incident and increases the numbers of police transportation to hospital while reducing arrests (Paton et al., 2016). However, there has been a lack of rigorous evaluation and understanding of its exact effectiveness to inform wider policy or practice (Forchuk et al., 2010; Paton et al., 2016).

In a study assessing what is needed for police in Scotland with regards to mental health, McKinnon (2014) highlighted that improving police training on mental health in line with physical health is a minimum requirement. Training should help officers identify mental illness or distress and divert the individual appropriately (Wood and Beierschmitt, 2014), thus preventing unnecessary arrests.

Although police training is encouraged in most interventions, the amount of training and a high level of understanding of its purpose is important (Martínez, 2010). A study of training interventions across Ontario found that 27 out of the 31 approaches surveyed provided five hours of training or less when the recommended amount of training was 40 hours (Durbin et al., 2010). Moreover, Ogloff et al. (2013) administered three different mental health screening tests to determine the number of people in custody who required specific mental health care. The results varied depending on the test administered which indicates that there can be an inconsistency in identifying mental health issues even if a standardised test is used. Therefore, it is important to note that police training is not proposed as a replacement for professional mental health assessments but rather should ensure police can identify and divert people in distress appropriately.

Despite this, however, the Ontario study also found that training helped police to identify mental health issues but they encounter a problem when trying to get support from mental health services. Therefore an intervention which falls into this category of increased training would work most effectively when it also provided increased liaison with mental health staff. Otherwise, police officers use of training and their ability to adopt it to tackle mental health issues is highly reliant on their local knowledge of health services and the individual needs of the person in distress (Wood and Watson, 2017).

Crisis Intervention Teams

Crisis Intervention Team ( CIT) models provide increased training and are particularly popular in America. Police officers are specially trained by mental health professionals to improve the identification of mental health issues and knowledge of care pathways available (Blevins et al, 2014). Some models also have non-refusal drop off points where police can take a person in distress and their care is guaranteed to be taken over by health services. The key aim is to improve the ability for officers to resolve the incident at the scene through the use of their training or with appropriate referrals (ibid).

Wood and Beierschmitt (2014) analysed mental health calls in one area of Philadelphia over eight years and identified that repeat callers to the CIT service created mental health hotspots. These repeat incidents highlight that CIT is helping police identify mental health issues more appropriately but the lack of long-term support for those who are in distress mean that the number of incidents is not reducing.

Existing studies, particularly outcome evaluations, on CIT are limited (Blevins et al., 2014). Evidence reviews are not available in this area because official data are not consistently recorded; the questions of interest are not covered in standard concern forms; and a lack of data sharing procedures means that a call from a person in distress cannot be followed through the system to gather information on the concern and the outcome (ibid).

Mental Health Intervention Teams in Australia

Another example of an increased training intervention are the Mental Health Intervention Teams ( MHIT) in Australia. The evaluation of MHIT measures the intervention's success against its initial aims (Herrington and Pope, 2013). It found:

• The length of time spent by police at mental health incidents reduced overall. This reduction was put down to a smoother hand-over to health staff due to increased knowledge of symptoms and health terminology.
• The roles and responsibilities of each service were reinforced during the pilot which reduced unnecessary police attendance at incidents.
• There was improved collaboration between the services, however, there were still tensions around the reduction of police burden causing an increased health burden. This was recommended to be addressed by use of a centrally funded, holistic approach towards mental health to reduce competition between the services for resources.
• There was no significant impact in terms of reducing incidents or injury on mental health cases. Still, it was noted that previous cases of injury were small and often in these cases injury had been self-inflicted as a result of self-harm. Training is unlikely to reduce the cases of self-harm as they tend to occur before an officer has attended,

This evaluation indicates that the training has improved collaboration and confidence while reducing the burden on police time however, a more rigorous evaluation with comparators would help to support the conclusions.

Connect Project

A systematic review on police training programmes, conducted as part of a wider study by the University of York and North Yorkshire Police, provided some valuable insights into the use of police training interventions in the UK. It has to be noted that some of the conclusions were difficult to draw based on the wide variation of study designs and low quality of studies identified in the systematic review. This is partly due to the lack of studies which focused on training programmes specific to police and so the scope of articles included in the systematic review had to be widened to include other non-mental health trained professionals (Booth et al., 2017). Nevertheless, of the studies that were available with a specific mental health focus, a statistically significant improvement was found on the confidence and attitudes of trainees during mental health interactions. However, when looking at training that was provided only to police officers, there was no statistically significant impact on either skills or arrest rates. However, the study did find that there were positive effects on training when it included a dramatization or role-play element and when it was delivered by mental health staff.

On the back of this systematic review, the Connect Project developed a training package delivered to police by mental health practitioners. To determine the effectiveness of this package, a randomised control trial ( RCT) was conducted of 12 police stations, half of which received the training and half had not (Scantlebury et al., 2017). After six months the study found that the training did not have a significant impact on the reduction of incidents reported to the control team, however, it did improve how police record the incidents. This supports the premise that increased training leads to change in attitudes and confidence of officers.

Moreover, the systematic review and RCT supports literature which promotes joint training delivered by mental health staff (Fenge et al., 2014; Kesic et al., 2013; Wood and Watson, 2017). This provides insights into the roles and responsibilities of the health service for police officers and vice versa which could challenge professional cultural barriers in place between the two services with regards to this area.

Alongside opening up a dialogue between the services on the front line, the Connect Project identified the need for there to be a high level of co-ordination between the services for interventions to be successful (Solar and Smith, 2016). Currently, front line decisions are made by services independently and locally by regional boards, quasi-governmental agencies, and policy programmes (ibid). Having one police force in Scotland centralises this more so than in England but co-ordinating with health boards and local authorities has the potential to be relatively complex. In the face of budget cuts and austerity this is likely to create an even greater degree of complexity as neither service has the resources to take on the co-ordinating responsibility (ibid).

Summary

The evaluations of interventions that adopted an increased training model have found that:

• Police training on mental health is essential for any collaborative intervention to ensure police are able to firstly identify people in distress
• Training reduces unnecessary arrests and hospitalisation, provides smoother hand-overs and, ultimately, reduces the time police spent on incidents
• Training improves police confidence when responding to mental health calls
• Joint training with elements of role playing is an effective approach
• There must be adequate resources to deliver the required level of training
• An element of liaison must run parallel with increased training to ensure care pathways for service users are improved alongside improved identification
• Inconsistent recording has reduced the quality and comparability of evaluations.

Embedded Mental Health Staff in the Police

Another category of intervention is the embedding of a mental health professional within an area of the police, usually in a contact control room ( CCR) or custody suites.

One of the key benefits on having a nurse within police settings is that they have access to both health and police records and information. Moreover, closer working and communication should improve information sharing and collaborative working. Wherever a member of mental health staff member is employed, the aim is to provide police officers with a direct contact to gain advice, support or on occasion provide a mental health assessment.

Interventions include:

• Some models of Street Triage
• Liaison and diversion
• Embedded staff in CCR

Street Triage

The traditional design of Street Triage aligns better with the co-response team category – to be discussed in more depth in the next section. However, two Street Triage pilots fall under the 'embedded mental health staff member' category. Firstly, the London Street Triage pilot was conducted across four London boroughs and provided 24 hour mental health telephone assistance to police (Hobson et al., 2015).

The evaluation of the London pilot (Hobson et al., 2015) found that:

• There was an increase of mental health incidents across the pilot areas
• The use of police custody as a place of safety declined
• While uptake of the service was initially low, police who did use the service tended to use it for place of safety recommendations

The findings, however, were not significant as similar trends occurred across London where no pilot was implemented. Also, the pilot was not fully staffed which was attributed to the low uptake as knowledge of the service and its availability was not as high as anticipated. Moreover, although police custody reduced as a place of safety this could be because officer who were utilising the service were calling for place of safety recommendations. Although this is a positive outcome it is not utilising the service to its full potential and further police training would be necessary to ensure police fully understand the purpose and remit of the scheme.

The second relevant pilot was conducted across nine areas in England and, although some of the pilot areas adopted a co-response model, most embedded a mental health nurse within a police setting or provided a dedicated phone-line for police to contact a nurse directly.The evaluation (Reveruzzi and Pilling, 2016) found that effective outcomes of the pilot included:

• the joint training programmes including clarity about the population to be served and agreed referral pathways
• joint ownership at a senior management level
• co-location of police and health staff such as in a control room or the provision of a dedicated phone line
• development of agreed protocols
• regular review process

A key finding is that much of the success was attributed to increased communication between the services and having a dedicated mental health nurse to provide support and information. Although it is difficult to differentiate between the pilot areas which adopted a co-response model, much of the data collected, regardless of the model, referred specifically to the above outcomes which are associated with having a nurse embedded in police settings.

Liaison and Diversion

Liaison and Diversion is another intervention which falls under this category. Maskrey et al. (2016) define diversion as the removal of a person suffering from mental ill health who is in contact with the criminal justice system to a more appropriate part of the criminal justice system or to community care in order to provide better outcomes. Liaison and Diversion (L&D) approaches aim to provide an assessment at the earliest point of contact with the criminal justice system so the individual can be diverted where possible or appropriately supported throughout their journey in the criminal justice system (Kane et al., 2017). They were first recommended in 1992 but have not always been consistently applied because of poor implementation, under funding and a lack of cross-stakeholder understanding (Fenge et al., 2014). The approach has since been endorsed for a national roll out in the 2009 Bradley review however its implementation at a national level is still inconsistent.

A refreshed L&D model was trialled across ten areas in England in 2014 and evaluated in 2016 (Disley et al., 2016). The evaluation found that:

• support workers who provide care after an individual has been referred through L&D are a key strength to ensure future engagement
• information sharing was recognised as a strength of the scheme
• valued partnerships are essential to ensure that referrals are appropriate
• across the ten test sites, relationships varied but consistently police were found to respond to the intervention well in all areas
• embedding a nurse in a custody suite was more expensive than prior approaches but it addressed more mental health cases and, as a result, the cost per case was less

Embedded Staff Member in Contact Control Room ( CCR)

The approach of embedding a mental health nurse in a control role was piloted and evaluated in Norfolk Constabulary, England (Maskrey et al., 2016). The method seeks to reduce demand on police officers, reduce risk to harm to all parties and ensure appropriate care is provided by providing early intervention before a crisis point is reached (Gwent Police, 2016).

The evaluation of the Norfolk pilot found that the intervention reduced the use of s136 apprehensions, however, as an observational study no control group was used to measure the extent (Maskrey et al., 2016). The lack of a control group also meant that time and cost savings could only be inferred. From the estimates it is likely that the intervention reduced the cost to police and health services by reducing s136 detentions, however, even the estimated savings would not cover the cost of implementing the intervention. Still, further quantitative and qualitative data collected in the evaluation found that the intervention successfully supported mental health and police staff while providing better outcomes for service users. These benefits must also be considered alongside cost effectiveness. The evaluation also found that there is scope within the intervention's current approach to expand its involvement in cases of self-harm, vulnerable adults and detentions under the Act.

The Norfolk model had four mental health nurses in the control room during peak hours to support control room staff and provide information and advice to police on scene or a person in distress. This was deemed more suitable than a street triage approach due to the rural structure of Norfolk and could be considered suitable for the rural areas of Scotland.

Information sharing was a key aspect to the success of these approaches (Fenge et al., 2014). The Norfolk pilot had written an information sharing agreement to be signed by relevant partners to ensure this. However, at the time of the evaluation the NHS trusts involved had yet to sign which, therefore, could have inhibited the accuracy of the evaluation or the success of the pilot (Maskrey et al., 2016).

Summary

In summary, evaluations of embedding a mental health member of staff in the police show that:

• Training is needed to ensure that police are utilising the service at the right times and for the right reasons.
• Information sharing between both services improves with this model which improves communication overall.
• Offering a telephone consultation works well as it provides support to more people than if the same nurse had to travel to conduct a face-to-face assessment. This has obvious benefits in rural areas and can reduce further distress caused by police or ambulance transportation.
• The cost of the service appears to exceed the potential cost savings however as it reaches more people the cost per case is less.
• A lack of comparative studies means that findings, particularly cost savings, can only be inferred.

Co-Response Teams

Co-response teams are made up of a mental health professional, often a nurse, and a specially trained police officer. The design of the co-response models vary depending on the intervention and the geographical location in which it is based. The differences are exemplified in the Street Triage pilots in England in which one had a mental health nurse on patrol with a police officer while another had a recovery programme integrated into the police response (Horspool et al., 2016). Co-response models that have been adopted and evaluated internationally include:

• Some Street Triage models
• Mobile Crisis Response Units
• PACER

Co-response models in general have been studied with more use of comparison or control groups than other models. A study conducted in Canada compared an area with a co-response team in place to a control area with no intervention before implementation, after one year, and after two years. The study found that, although the use of the co-response model increased and was in high demand, the call time reduced significantly (Kisely et al., 2010). The study assessed quantitative data of out-patient attendance after the intervention and qualitative data of service user experience and concluded that the intervention provided increased engagement and positive service user experience. However, the study did have limitations, particularly with recording bias and difficulty in generalising the qualitative findings, but it is one of the few studies with a control group and mixed methods evaluation.

A further study of service users experiences of a co-response model compared to police only responses found that co-response teams were more likely to provide voluntary and mandated escorts while police only models provided more involuntary escorts (Lamanna et al., 2017). Focusing on the service user experience, Lamanna et al. (2017) also found that the users commented on the knowledge, empathy and communication skills of nurses in the co-responding teams which improved de-escalation of the crisis. Similarly, a review of joint response models found that they referred many more people to mental health services, however, this provided a less traumatic and more engaging outcome for the service user (Shapiro et al., 2015).

Co-response models demonstrate an improved co-operation between services however a lack of resources is hindering this co-operation as the services, particularly health, are ill equipped to deal with this new demand (Morgan and Paterson, 2017). A study conducted by Ogloff et al. (2013) demonstrated the impact of this by surveying police officers in Australia. They found that 55% of officers surveyed would ideally seek the help of a Mental Health Crisis Assessment Team but, due to a lack of availability of the service, would actually detain the distressed individual under their equivalent of the Mental Health Act. Similarly, in a study comparing three different types of interventions in Canada, Durbin et al. (2010) found that, even when co-response models were in place, their use was low with the majority of officers opting to use it in less than 25% of cases.

Street Triage

In terms of specific co-response interventions and their success, much of the research is focused on Street Triage. The Nottingham Street Triage scheme provided two cars with a police officer and a mental health nurse available during the peak times of 16:00- 01:00 ( NHS England, online). The one year pilot saw a 53% reduction on s136 detentions and, although there was a liaison service already in place, its success is contributed to the collaborative approach so much so that the funding for the scheme has continued beyond the pilot.

A Street Triage pilot in nine areas across England, identified that all areas had to adopt a different design as a result of varying geography, population and level of mental health detentions (Reveruzzi and Pilling, 2016). An evaluation of the pilot found that it reduced s136 detentions, increased the use of health based places of safety, and reduced time spent in police custody (ibid). However, as there were a number of different approaches and models adopted across the pilot areas, it is difficult to infer which model contributed to the positive outcomes. Furthermore, although the pilots were all implemented locally and were run individually by selected police forces, they often ran across two or more NHS Trusts (Horspool et al., 2016). Mental health staff involved in Street Triage pilots, therefore, found it difficult to collaborate between the different health trusts and between institutional and community services. The success of the interventions was largely based on their own knowledge of the operating location and the services available (ibid). Therefore, increased knowledge of the services is necessary on both sides for the co-response model to be successful or it may be more appropriate to implement interventions more locally.

PACER

The Police Ambulance Crisis Emergency Response ( PACER) model in Australia, and subsequent adaptations such as A- PACER and N- PACER, were developed from CIT and other approaches that have been successful across America (Huppert and Griffiths, 2015). PACER was implemented alongside increased police training and saw mental health professionals travel with police officers and respond to mental health calls or provide telephone consultations when a face to face response was not possible. This would ensure an appropriate and timely assessment by a mental health professional and more accurate and quicker hand-overs. The police officer can then manage the safety of the situation and will only provide transportation to an emergency department or a mental health hospital if the service user presents as violent.

A study of the A- PACER model compared to other Australian approaches found that police were less forceful, mental health staff were quicker to respond and assessed crises more accurately in person than over the phone (Evangelista et al., 2016). This resulted in less people being inappropriately imprisoned or hospitalised. An assessment of N- PACER model in Australia found that this approach provided service users with a greater perception of procedural justice, however, it was still found to be coercive and reduce service user autonomy (Furness et al., 2016). Moreover, although service users tended to be more satisfied with the co-response model some felt it would be more appropriate for health staff to respond alone (Evangelista et al., 2016).

A thorough assessment of the PACER pilot was evaluated against three outcomes: length of stay in emergency departments; time spent at crisis by the first responder; and cost effectiveness. Based on these outcomes it was found (Allen Consulting Group, 2012) that:

• Time in emergency departments was reduced due to improved handovers
• Police as first responders spent less time on mental health incidents due to less time spent in emergency departments and reduced need for police transport
• PACER appears to be less expensive than time and resources spent on alternative approaches. This costing, however, is based on many assumptions.

Moreover, the PACER approach receives strong ownership from both health and police sides but police refer to the success of the intervention being down to the locally devolved structure of the police force which contrasts with the centralised decision making structure of the Department of Health in Victoria (Allen Consulting Group, 2012).

Summary

Co-response models, particularly Street Triage and PACER pilots, have generally received positive evaluations. Findings include:

• Individuals who are in distress and are seen by a co-response team are less likely to be hospitalised, detained or charged unnecessarily.
• Co-response models did not provide as much geographical coverage or respond to as many cases as increased training models but they provided a better resolution on the scene on the incident.
• Service users generally had an improved experience and better onwards engagement with a co-response model to the extent that police contribution is deemed unnecessary.
• The local implementation has been identified as a key to the interventions success as it improves knowledge of staff and co-ordination. Such interventions may therefore be more strategically placed in local areas but must also balance the standardisation of a national approach.
• Although the interventions are recognised to be financially and resource intensive this is deemed as a short term impact. This investment is seen to provide further long term benefits as the skills learned during the co-response team hours can be applied by police on general policing duties. Moreover, working together would improve understanding from both health and police sides of the other sides' responsibilities and pressures which would further challenge cultural barriers.