Systematic reviews
Overall, 31 review articles evaluating PRMHT interventions were identified. This study focused on those reviews that used systematic, transparent or reproducible methods, as determined by the DARE criteria.31 This resulted in five reviews being included in the synthesis.11,12,14,15,39,40 There were six articles published for the five reviews as Kane et al.11 also published their review protocol. The remaining 25 review articles were excluded from the synthesis as their methods were deemed less robust (according to DARE criteria) or were not reported. These reviews are listed in Appendix 5 and DARE assessments for all identified reviews are presented in Appendix 6.
Below is a summary of the five included systematic reviews. Three reviews were evaluations of PRMHT interventions,11,12,15,40 one review evaluated mental health training programmes for PRMHT39 and one was a scoping review of interagency collaboration models for PRMHT.14 Details of the study characteristics are also provided in . The level of detail of participants, interventions, comparators, outcomes and findings varied between reviews. Terminology also differed and was often undefined. We have endeavoured to report the findings in as much detail as was available from the original reviews.
Characteristics of systematic reviews
Characteristics of systematic reviews
The reviews of PRMHT effects included studies conducted in Australia, Canada, the Netherlands, Sweden, the USA and the UK, published between 1999 and 2016 (when reported).
Study designs of the included primary studies varied across the reviews. The reviews of PRMHT interventions included a meta-analysis, controlled studies, quasi-experimental controlled studies, pre–post comparisons and descriptive studies. The review of training programmes for PRMHT included a systematic review, RCTs and prospective non-RCTs. Most studies in the scoping review of interagency collaboration models were descriptive but this review also included mixed-methods studies, service evaluations, scoping reviews and observational studies.
Police-related mental health triage interventions
Details of the interventions of PRMHT were generally only briefly reported in the reviews and included the following.
Pre-arrest diversion
Control room call-handlers identify incidents when people are in a mental health crisis and to which they can despatch unaccompanied police officers with special mental health training who can then refer to mental health services [e.g. CIT or a Mental Health Intervention Team (MHIT)].11,12,14,15,39,40
Co-response
Co-response includes front-line police officers being supported by a MHP as a joint, on-scene response (hours may be restricted, i.e. co-response with MHPs may be operated only during night shifts) and/or a MHP in a control room [e.g. street triage teams in the UK11,12,14,15 or a police, ambulance and clinical early response (PACER) in Australia].12 The Parker et al.14 scoping review described interventions in which the MHP exclusively provided advice from a separate location (e.g. via telephone) as befitting a ‘consultation’ model.
Service integration models
This intervention was described in the scoping review of models of interagency collaborations.14 The model describes multiagency integrated services to create a network to bridge gaps between services, decrease arrest, decrease violence, improve educational attendance and completion, and reduce symptoms of mental ill health and psychological distress. These may typically involve a network co-ordinator (often a MHP), a personalised action plan for individuals in mental distress and active follow-up with signposting to the relevant agency. This is somewhat broader than the focus of the current review as it encompasses far more than triage and/or crisis work.
Training programmes and resources
Training and resources were described for the various PRMHT interventions, which had either a specific or a broad mental health focus.39
Mental health triage providers and users
Whenever reported, the providers of the PRMHT interventions (service providers) were police officers,11,12,14,15,40 community service officers15 and MHPs.15 The scoping review by Parker et al.14 reported a wide range of relevant agency collaborators from health, welfare and social care services. The review by Booth et al.39 evaluated training programmes and described a range of trainers, such as police officers, MHPs, educators and service users. Trainees included police officers as well as others from areas such as education, welfare and social care.39
People coming into contact with PRMHT interventions (service users) were described in only two of the reviews. Taheri40 evaluated CIT interventions and broadly described service users as those diagnosed with mental ill health. The scoping review by Parker et al.14 described service users as adults, children, young people and mixed populations. No further details on participants were reported in any of the reviews.
Outcomes
Organisational- and service-level outcomes were the most frequently reported across the reviews. These included rates of detention,11,12,15 use of HBPOS,11,12 arrest rates,40 mental health referrals, police officer time dealing with events and police officer safety.11,12,15
The review of training by Booth et al.39 reported on effects on police officer awareness, attitudes, beliefs and knowledge as a result of mental health training. A range of outcomes was also reported for the training of those in education, welfare and social care including awareness, knowledge, self-efficacy and changes to the environment.39
Approach to synthesis
Only the review by Taheri40 conducted a meta-analysis. The remaining review authors conducted narrative syntheses because of the diversity of the primary studies and PRMHT interventions.
Overlap of primary studies within the review
There was some overlap of primary studies among the reviews themselves, and also in the additional primary studies included. The overlapping studies identified are listed in .
A summary assessment of the overlap indicates that a number of studies appeared in one or more of the effectiveness reviews.11,12,15,40 The review by Taheri40 was also included in the Kane et al.11,12 review. Studies in the review of training by Booth et al.39 appeared in some of the effectiveness reviews. Some of the studies in the scoping review by Parker et al.14 were also included in the effectiveness reviews, but findings were not reported.39
Although there was some evidence of overlap of primary studies between the reviews, it is interesting that there was not a greater overlap given that all the reviews met our inclusion criteria. This perhaps illustrates the variety in forms of PRMHT interventions that have been evaluated and the lack of consistent nomenclature in this area in general.
Quality of the systematic reviews
The reviews were assessed for risk of bias using the Egan et al. criteria.32 Most had some potential for error and bias in the review process. provides details for each review.
Quality assessment of systematic reviews
Booth et al.39 fulfilled all the criteria. Paton et al.15 met all the criteria but did not report if more than one person was involved at each stage of the review process, which means that there is potential for reviewer error and bias in the selection of studies. Kane et al.11,12 met some of the criteria but despite reporting the intention to assess the risk of bias of included studies, the results of this assessment were not reported and did not appear to inform the findings. The authors also did not clearly report the study designs or number of included studies. These omissions mean that it is difficult to judge the reliability of the individual studies. The involvement of more than one reviewer was reported for only some parts of the review process, meaning that there is the potential for reviewer error and bias. Taheri40 was the sole author, so there is the potential for error and bias in the selection of studies, data extraction and synthesis. Risk of bias of the primary studies was not assessed, meaning that it is not possible to judge the reliability of the results from the meta-analysis.
The scoping review by Parker et al.14 met all the criteria apart from assessment of primary studies, which is acceptable for a scoping review that does not report effectiveness findings.14
Findings from the reviews
We have focused on the outcomes relevant to our review questions. Two of the included reviews had broader objectives and we have only used the sections relevant to PRMHT interventions.14,15 A summary of the relevant findings are reported in .
Summary of results of systematic reviews
The authors of the reviews reported on variations of PRMHT interventions in primary studies of various designs. It is, therefore, not possible to group the studies by model of PRMHT or outcome; the findings for each review are reported separately below.
Kane et al. (2017)
The most recently published systematic review, by Kane et al.,11,12 included studies that reported on-street triage and CIT, plus one study that compared three different approaches with police involvement.
As a result of street triage co-response interventions, a quicker and more appropriate response by the teams was reported overall. Reductions were reported in single studies for formal detentions, an increased use of HBPOS and a reduction in time spent on-scene by the team.
Kane et al.11,12 also reported findings for comparisons between CIT-trained officers and non-trained officers. Trained officers were more likely than non-trained officers to direct people with mental ill health to mental health services, but this depended on local services available. Trained officers also demonstrated different approaches to people with mental ill health. Although it should be noted that police officers self-selected for training, that might indicate an existing difference between trained and non-trained officers, which could have an impact on approaches and attitudes.
One study in Kane et al.11,12 reported on a comparison of three different models operating in the USA, including civilian police employees with additional training to assist police officers in mental health emergencies, a CIT unit and a mobile crisis unit. Authors reported that ‘the dispositions of cases handled by the specialised response personnel were found to be related to the programme type’12 (no further details reported), although all three models reported an average rate of arrests of 7%. Factors reported to be related to success were the existence of a psychiatric triage or drop-off centre to which the police could transport individuals in crisis, and community partnerships in which police response is part of a wider response involving relevant agencies.
Taheri (2016)
Taheri40 reported that there were no statistically significant effects of CIT interventions, between CIT-trained and non-trained officers, on official or officer-reported accounts of arrests of people with mental ill health. However, the studies were heterogeneous and reported conflicting results.40
Paton et al. (2016)
The review by Paton et al.15 evaluated on-scene co-response or telephone triage by health or social care professionals in collaboration with police officers (some of whom were in receipt of CIT training).15 These included various collaborative partnerships. One UK study of a street triage intervention, in which officers received telephone triage, was associated with a reduction in S136 detentions and a number of calls were successfully converted to hospital admissions.42 There were also improvements in police time on the scene of a suicide call, and improvements in engagement with outpatient treatment by service users, although these findings were reported only in one or two studies.
Paton et al.15 also reported the effects of police officers receiving training in mental health. The interventions included CIT-trained officers and a MHIT. In one study, training seemed to result in higher levels of verbal engagement and negotiation, and more referrals to and transport for treatment. Lower arrest rates and reduced referrals to intensive psychiatric services were also reported in one study. In another study, a MHIT intervention found no significant differences in the use of force before and after training, and no perceived improvement in the quality of the relationship between officers receiving training and health-care staff. Trained officers spent less time dealing with events that related to the Mental Health Act 1983.2
Booth et al. (2017)
A review of mental health training of service providers reported findings across several interventions.39 The review found that CIT training may be effective in connecting people with mental ill health with appropriate psychiatric services via police officers. The training may also have had an effect on officers’ attitudes, knowledge and beliefs. However, the review authors state that these findings were based on limited- and poor-quality studies. No comparison groups were reported in the primary studies.
Education, awareness training and anti-stigma courses aimed at a range of trainees including police officers, MHPs, teachers, public health officials and voluntary sector staff reported various successes following training, including positive changes in awareness, knowledge and attitudes.
A study of the training for MHIT teams of front-line police officers in Australia found no substantial changes in practice or relationship quality compared with no training. There were also no significant differences in skills.
A number of other training programmes, not directly aimed at PRMHT, were also reported in the paper.
Parker et al. (2018)
The scoping review by Parker et al.14 did not report any findings from primary studies.
Primary studies
As the most recent search date in the included systematic reviews was 2016,12 the metasynthesis of systematic reviews was supplemented with primary studies published in 2016 or later. The aim was to consolidate the most recent evidence of the effectiveness of PRMHT interventions. As the primary studies provided greater detail about specific interventions, participants and results than the systematic reviews, these studies are discussed at greater length in this report. However, it should be borne in mind that these studies represent a much smaller body of evidence than the published systematic reviews summarised in Systematic reviews.
Overall, 20 primary studies were identified. The studies included PRMHT interventions from the USA, Australia and the UK.
We chose to focus on the UK studies because the context in which interventions are evaluated will vary by country. For example, there are important differences between the UK and the USA in terms of funding, delivery and governance of criminal justice and health services, as well as wider societal differences.
Although we focus our synthesis on UK studies, this report provides summary details of the other studies in Appendix 7.
Eight primary studies undertaken in the UK were identified in nine articles. One trial contributed two articles, including one on methodological and practical challenges of undertaking the trial. Articles were published between 2016 and 2017. All of the studies evaluated schemes in England. provides summary details of the study characteristics. Further details of the findings from each primary study are available in Appendix 8. The level of detail when reporting on participants, interventions, comparators, outcomes and findings varied between primary studies. Terminology also differed and was often undefined. We have endeavoured to report the findings in as much detail as was available from the primary studies.
Characteristics of primary quantitative studies
Characteristics of primary studies
The included studies were of varying designs. One RCT evaluated a training programme for front-line officers,47,48 although the remaining studies were largely descriptive, reporting little comparative data. Single studies used the following designs: a comparison of two different models of PRMHT interventions,44 a largely single-group evaluation during a PRMHT intervention with some comparison data from a cost–benefit analysis,46 an evaluation across multiple areas of PRMHT pilot interventions that reported comparative data only for some outcomes (e.g. S136 detentions),9 a descriptive study comparing data before and after an intervention,45 a pragmatic evaluation using a decision-analytic model exploring costs,43 a non-comparative assessment and evaluation of two areas implementing a PRMHT intervention,50 and a dissertation reporting data during the period of an intervention.49
One of these studies (Reveruzzi and Pilling)9 was also included in one of the reviews in the metasynthesis.12 Three studies that reported quantitative data9,46,50 also informed our synthesis of qualitative evidence (see Chapter 5).
Police-related mental health triage interventions
Eight studies (in nine articles) evaluated PRMHT interventions.9,43–50 Most studies evaluated variations of co-response or consultation models that were often referred to as street triage. The intervention characteristics varied across the studies, but all could be broadly classified in terms of the primary role of MHPs (). These fell into the following categories.
Characteristics of PRMHT interventions in primary studies reporting quantitative data
Joint on-scene response with police officers (five studies)
Mental health professionals were partnered with police officers to jointly attend incidents. Schemes varied across the studies, with police officers attending in uniform and teams arriving in marked or unmarked police vehicles or community vehicles. Hours of operation varied and were often limited to ‘peak response’ hours. Outside the hours of street triage operation, ‘usual response’ from police officers and access to mental health services continued.9,43,45,46 One study reported a telephone line also being made available during peak hours to provide support to officers.46 In one study, joint on-scene attendance of MHPs and police officers in one county in England was compared with telephone support only from a mental health nurse in another county.44 In this intervention, a police unit would be on-scene before the street triage team arrive to assess the risk and confirm the need for the street triage car. This is to ensure that risk assessments are carried out before MHPs arrive at the scene, as police have a duty of care towards colleagues who have not had the same degree of training in conflict management, and who do not have the clothing and/or equipment to manage such situations. The studies covered a number of areas across England including parts of Derbyshire, Essex, Northumberland, Tyne and Wear, Norfolk, Suffolk, Sussex, Thames Valley and the West Midlands.
Telephone support alone (one study)
A model in which telephone support was provided by a MHP from a police control room was employed by the British Transport Police (BTP) service, which covers large parts of England across many health trust and police service boundaries.9 The study by Jenkins et al.44 compared the joint on-scene response of MHP and police officers in Ipswich, Suffolk, with telephone support only from a mental health nurse in Norwich, Norfolk.
Joint on-scene response and telephone support (one study)
A dissertation by Brace49 evaluated a street triage scheme in England (area not reported) that consisted of a MHP working in the police control room as well as a joint MHP and police officer on-scene response, both overnight. The aim was to provide an out-of-hours service because during office hours a service is provided by police officers and the mental health crisis team.
Telephone advice to on-scene officers and attend if necessary (two studies)
Two studies9,50 used this model of intervention in which MHPs did not routinely attend incidents but were available to do so if required. One study, covering part of Northamptonshire, was not clearly reported in terms of the intervention but appeared to involve police officers, a MHP and control room staff over certain periods of operation each week (details not reported).50 The other study covered parts of Devon and Cornwall, where the MHP was based in the control room, and also the area covered by the Metropolitan Police Service, where the MHP was based in a mental health trust.9
Telephone and/or on-scene response without police presence (one study)
A telephone and/or on-scene response without police presence was reported in two pilot evaluations. MHPs were situated in a local hospital (North Yorkshire) or a Crisis Assessment Service (West Yorkshire).9
Training programme for police officers (one study)
The RCT by Scantlebury et al.47 evaluated a training programme conducted in North Yorkshire that was aimed at front-line police officers who come into contact with people with mental health problems. A qualified and experienced MHP delivered the 1-day face-to-face mental health training in addition to routine police training.
Mental health triage providers and users
There was a lack of information regarding participant characteristics.
Brace49 reported service providers as police officers with the rank of Constable or Sergeant, who had regular interactions with people in a mental health crisis. The remaining studies did not report any details of service providers of street triage interventions.9,43–46,50
Data on service user characteristics were similarly limited. Reveruzzi and Pilling9 reported the age of service users who came into contact with street triage teams in pilot areas across England, with 29% of service users aged between 18 and 30 years, and over half (52%) of service users aged between 30 and 65 years. Jenkins et al.44 reported no significant differences between service users for age or gender before or after a street triage intervention, and that most service users were of a white British background. The remaining studies did not report any data on service users who came into contact with street triage teams.43,45,46,49,50
Three studies reported some limited population data about areas of operations of the street triage schemes that were largely figures often gathered from National Census data.9,43,44
The RCT on the training of officers reported trainers as qualified and experienced MHPs and trainees as front-line officers of the rank of Constable, Sergeant, Inspector and Police Community Support Officers.47,48
Outcomes
Outcomes varied widely across the included primary studies. Very few outcomes were evaluated using concurrent comparative data. Most outcomes were collected at an organisational or service level. All authors reported using routinely collected data from police and health trust records.
Outcomes included:
national trends across England
numbers and rates of S136 detentions
number of assessments taking place in police stations
number of assessments/detentions taking place in hospital
number of admissions to hospital as a result of S136 assessment
contact with community mental health teams or follow-up services
costs.
The RCT of a bespoke training programme for police officers reported on:
the number of incidents reported to a control room resulting in a police response
the likelihood of incidents having a mental health tag applied (i.e. a tag applied by police officers to indicate that mental health is a factor in an incident)
the number of individuals with a mental health warning marker involved in any incident with a police response (i.e. a marker applied by police officers to an individual’s record to indicate they have mental health problems).
Many studies recorded the above outcomes during the period of the street triage interventions but without any concurrent comparative data.
Quality assessment of the primary studies
The RCT by Scantlebury et al.47,48 was rated as having a low risk of bias for most criteria. However, there may have been the potential for some contamination between intervention and control sites, and it was unclear whether or not incomplete outcome data were adequately addressed. Details are provided in .
Quality assessment of a RCT
The remaining primary studies were at risk of multiple biases because of their designs and a lack of reporting of methods, which undermined the reliability of their findings. A summary of individual study methods is reported below.
The study by Heslin et al.43 used routinely collected data on S136 detentions before and after a street triage intervention. A number of the data were available only for 4 months, so authors extrapolated these data for the 6-month evaluation period; a cost-offset analysis was also conducted.
Jenkins et al.44 compared two separate models of street triage in different areas using routine data from police and health trust records, collected retrospectively at time points before and after each intervention.
The study by Keown et al.45 used routine data collected retrospectively from police and health trust records before and after a street triage intervention.
A report by Reveruzzi and Pilling9 collected data across nine pilot areas of street triage in England. Data were collected prospectively across all areas by either police or health staff, depending on the area. However, data were not collected consistently across the pilot areas. The only comparative data provided were for S136 detentions, for which there was a high proportion of missing data in some areas.
A report by Senker and Scott46 was a single-group evaluation of both a pilot period and a full street triage service. It is unclear whether data were collected prospectively or retrospectively. The only comparative data reported were for the cost–benefit analysis.
One report50 and a dissertation49 provided no comparative data at all and are not included in the synthesis of findings.
Findings from the primary studies
To be able to answer the research question regarding the evidence of the effectiveness of the PRMHT models, we focused on comparative data. Two studies provided a ‘snapshot’ of the schemes, with only cross-sectional data collected during the operation of a scheme or pilot study, with no comparison data from before the intervention or from another model.49,50 Although these data are interesting, it is not possible to determine the effectiveness of the interventions they describe. These studies are, therefore, not included in the findings below.
The text below summarises the findings of the primary studies. Appendix 8 provides further details.
Number and rates of section 136 detentions under the Mental Health Act 1983 (four studies)
It should be noted that some authors indicated that these data were gathered at the incident level, not the individual level. Therefore, one individual may have received more than one S136 detention.
Reveruzzi and Pilling9 evaluated the nine pilot areas of various models of street triage implemented in England. Street triage schemes in the West Midlands, Derbyshire and Thames Valley (all joint co-response schemes) saw the largest reductions in S136 detentions, with decreases of 27.5%, 25.3% and 22.7%, respectively. Significant reductions in S136 detentions were also found in West Yorkshire of 19.8% (MHP telephone and/or on-scene response without police presence), in Sussex of 18.3% (joint on-scene co-response scheme) and in Devon and Cornwall of 15.5% (MHP telephone advice to on-scene officers). The Metropolitan Police Service (MHP telephone advice to officers and attend if necessary) and North Yorkshire (MHP telephone and/or on-scene response without police presence) increased S136 detentions by 15.1% and 19.4%, respectively. BTP (MHP telephone support only) results were not included in these figures.
There were significantly fewer S136 detentions after the introduction of the street triage intervention (joint on-scene co-response) in Eastbourne and Sussex than before the intervention.43 However, the authors noted that in only 6% of cases people were detained under S136 by street triage teams, whereas the remaining cases were made by officers during shifts when the street triage team was unavailable. The authors also noted that across the rest of Sussex S136 detentions increased, although this was not statistically significant. This study evaluates the area of Eastbourne, an area that was also included in Reveruzzi and Pilling9 as reported above, although some figures differ. This may be because each study referred to different years of census data to calculate some findings.
Two street triage schemes were compared by Jenkins et al.44 The team in Ipswich, Suffolk, was a joint on-scene co-response team intervention, compared with a MHP telephone support-only scheme in Norwich, Norfolk.44 Over the total evaluation time period (pre and post intervention), the telephone support team had statistically significantly fewer S136 assessments per 100,000 people than the joint on-scene co-response team (p = 0.01). The difference was statistically significantly greater between the areas in the pre-intervention period (p < 0.01). Post intervention there was a small, non-significant increase in those detained under S136 in the telephone support-only area but a large reduction in S136 assessments in the co-response team area (p = 0.01).
There was a 60% reduction in the rate of S136 detentions in the first year of a street triage scheme (joint on-scene co-response) reported by Keown et al.45 in areas of Northumberland, Tyne and Wear. The greater the rate of street triage in an area, the greater the reduction in the rate of S136 detentions in the same area (p = 0.003). Data across three of the areas (Gateshead, South Tyneside and Sunderland) that first introduced a street triage scheme showed a statistically significant progressive reduction in the number of S136 detentions during each subsequent 3-month period in the first year of the street triage scheme compared with the previous year. This resulted in a 65% reduction in the first 3 months, 73% reduction at 6 months, 83% reduction at 9 months and 88% reduction at 12 months (p = 0.001). Additional data for Sunderland were available and authors reported a 78% average monthly reduction in S136 assessments following the introduction of street triage.
Number of assessments/detentions taking place in police stations (two studies)
Keown et al.45 reported that only a small number of assessments were conducted in police stations both before and after the introduction of a street triage scheme (joint on-scene co-response) in one area of the intervention (Sunderland).
In Eastbourne, Sussex, there were significantly fewer detentions in custody after the introduction of a street triage (joint on-scene co-response) than the usual response before street triage (p < 0.05), as reported by Heslin et al.43 Although because the street triage hours of operation were limited, a number of the detentions in custody during this period were by usual response.
Number of assessments in hospital (one study)
Keown et al.45 reported that most S136 assessments were conducted in a dedicated S136 suite in a psychiatric hospital in Sunderland (just one area reported in the study), as they had been before the introduction of a street triage scheme (joint on-scene response). There were also no significant changes in the proportion of individuals detained in hospital, voluntarily admitted or with no admission after the intervention.
Detentions taking place in hospital (one study)
There were fewer detentions in hospital after the introduction of a street triage scheme (joint co-response) in Eastbourne, Sussex (reported in Heslin et al.43), than before the intervention, but this difference was not statistically significant. Across the rest of the county, the number of people detained in hospital increased but this was not statistically significant. As reported above, because the street triage hours of operation were limited, a number of the detentions during this period were by usual response.
Number of admissions to hospital as a result of a section 136 assessment (one study)
There were no significant differences in admissions to hospital following S136 assessment reported in Jenkins et al.44 before the introduction of street triage schemes in Ipswich, Suffolk, and Norwich, Norfolk. After the implementation of street triage, there was a statistically significant higher conversion to admissions for a joint on-scene co-response in Ipswich than pre intervention (p = 0.01), but there were no significant differences for admissions before and after the telephone support-only scheme in Norwich. After the interventions, there was also a significantly higher admission rate following S136 assessment in Ipswich than in Norwich (p = 0.04).
The same study also reported that the proportion of individuals who were not admitted to hospital following a S136 assessment but had at least one subsequent S136 assessment in the following 4 weeks decreased in both street triage schemes, but the decrease was only statistically significant for the telephone support-only scheme in Norwich (p = 0.01 vs. p = 0.14). The authors reported that there were more individuals in Norwich who had some contact with community mental health services in the 2 weeks prior to a S136 assessment than there were in Ipswich, and that there were more individuals in Norwich who had some contact with community mental health services before the introduction of a street triage intervention as opposed to after this introduction than there were in Ipswich (p = 0.01). Whereas, in Ipswich, there were no statistically significant differences between contact before or after the street triage intervention. There was also no evidence to suggest that individuals assessed in Ipswich were more likely to have been deemed to have ‘no mental illness’44 before the introduction of street triage than after, but in Norwich there was evidence to the contrary (p < 0.01).
Contact with community mental health teams or follow-up services (one study)
There was some evidence in Jenkins et al.44 to suggest that if people in Ipswich were not admitted to hospital, they were more likely to be offered follow-up from secondary mental health services after the introduction of a street triage scheme (joint on-scene co-response) than before the intervention (p = 0.04). If follow-up was offered in Ipswich, there was evidence to suggest that the first follow-up appointment was more likely to be kept after the street triage scheme than before (p < 0.01). In Norwich, evidence from the street triage scheme (telephone support only) suggested that a person was more likely to be offered a follow-up before as opposed to after the intervention (p = 0.02), but there was no evidence of any difference between compliance rates before or after street triage implementation.
Costs (two studies)
Senker and Scott46 reported costs of a street triage scheme that had both a joint on-scene response unit and MHP telephone support in Essex. The authors reported potential annual benefits from street triage preventing the use of S136 and the wider impact of its use. These potential annual benefits included gross realisable savings for NHS trusts or Clinical Commissioning Groups of £347,200, with benefit to the value of £99,650 for the Essex police force in terms of reduced use of custody and reduced officer time attending S136 incidents, which realised a gross benefit value estimate of £446,850 and a net benefit of £179,758 when accounting for costs of running the street triage scheme.
Over the period of a street triage (joint on-scene co-response) evaluation of 26 weeks in Eastbourne by Heslin et al.,43 the total cost of the scheme was estimated to be £148,785. The street triage team attended a total of 233 incidents, giving an estimated unit cost of £630 per incident attended.
The study calculated average costs to payers resulting in £1043 per person for street triage and £1077 for usual response; CJS: £470 for street triage and £559 for usual response (representing a cost saving of £31,862 over a 6-month period); and NHS: £574 for street triage and £517 for usual response (an additional cost of £20,406 over a 6-month period to the NHS for the street triage arm).
Other outcomes (one study)
After a 1-day bespoke training programme for front-line officers, Scantlebury47,48 reported that there were no significant differences in the number of incidents requiring a police response between those police stations receiving officer training and those police stations that did not. There was also no significant difference between the intervention and control arm stations in the number of people with mental health warning markers on their record. However, there was evidence that incidents at stations randomised to the intervention arm were more likely to have a mental health tag (applied by officers to indicate that mental health is a factor in an incident) applied than incidents assigned to control arm stations (p = 0.001).
Metasynthesis
Which models have been described in the literature?
Models described in the systematic reviews:
pre-arrest diversion models (which include CIT)
co-response and/or consultation models (which include street triage)
service integration models (which include multiagency integrated services).
Models described in the primary studies:
What evidence is there on the effectiveness of these models?
There was a paucity of evidence on the effectiveness of PRMHT interventions from the included systematic reviews. Most of the outcomes were at an organisational or service level, with many of the data relating to S136 assessments or detentions. Overall, reductions in formal detentions and an increase in the use of HBPOS were reported, but results for other outcomes were mixed. Most of these results were based on the findings of just one or two primary studies of unknown or poor quality.
Review authors reported that few studies were of robust design, many were uncontrolled single-group studies often using retrospective, routinely collected data. Very few details of the context or content of the interventions were reported. Service users were described only generically. Authors of all of the reviews acknowledge the lack of robustness in much of the available evidence and have provided recommendations for future research.
Much of the evidence in the most recent primary studies relates to the MHP joint on-scene co-response with police officers. Although there were several reported reductions in S136 detentions, fewer detentions in hospital, higher admission rates and greater likelihood of being followed up by secondary mental health services (if not admitted), these results were not always consistent. Evidence on costs is limited and somewhat contradictory in relation to the possibility of costs shifting between agencies. However, these findings were largely from one or two studies with major methodological limitations that preclude strong inferences being drawn.
Few studies reported comparative data, making it difficult to determine the relative effectiveness of different models of PRMHT against each other or against usual practice.
Recommendations for research
Authors’ recommendations from the systematic reviews:
Rigorous, higher-quality evaluation is needed.
14,15Higher-quality evaluations are needed and should use self-report and official measures.
40Future evaluations should target health-related outcomes and have an impact on key stakeholders.
14Further build evidence to refine and develop integrated interventions.
11,12Longer-term follow-up of training programmes is needed, as is better-quality research evaluating training for UK police officers.
39
Authors’ recommendations from the primary studies:
Evaluate outcomes of street triage contacts and whether or not the schemes lead to improved patient outcomes, and delineate when street triage is the best option and when S136 might be a preferred pathway.
45Examine findings using a wider range of data, preferably prospective studies.
45Data-recording procedures should be reviewed and streamlined with clear expectations and accountabilities for all parties.
50Data should be collected on mental health and quality-of-life outcomes, including quality-adjusted life-years, to enable full assessment of cost-effectiveness and cost–utility of street triage.
43Capture data on the number and proportion of incidents that do not result in a detention under S136.
43Capture feedback from relevant professionals and individuals to enable a fuller understanding of the impact of the service.
44
