Consultation outcome

Police powers: pre-charge bail overview of the evidence (accessible version)

Updated 6 September 2021

Scarlett Furlong, Victoria Richardson and Andy Feist

January 2021

Acknowledgements

The authors are grateful to Dr Richard Martin (Assistant Professor of Law at the London School of Economics and Political Science), Professor Anthea Hucklesby (Professor of Criminal Justice and Head of the School of Social Policy at the University of Birmingham), and Professor Mandy Burton (Professor of Socio-Legal Studies at the University of Leicester), for peer reviewing this report.

Executive summary

Introduction

The police in England and Wales can grant pre-charge bail (PCB), also known as police bail, to individuals arrested on suspicion of a criminal offence but where there are no grounds to keep them in detention while the investigation continues. At this stage there is insufficient evidence to charge.

The main purposes of PCB can be summarised under three headings: − The protection of victims and witnesses, primarily linked to conditions applied to PCB such as no contact with the victim. − Investigative management, allowing investigations to progress to obtain evidence. − Suspect management, including reducing the risk of re-offending.

PCB was subjected to legislative change through the Policing and Crime Act (PCA) 2017. The changes were principally the result of concerns that suspects were spending extended periods of time on PCB, often for the case against that suspect not to be taken forward.

The PCA reforms – which created a presumption against the use of police bail – resulted in a precipitous drop in using PCB in the year after their introduction, mirrored by an increasing number of individuals being ‘released under investigation’ (RUI). The decreased use of PCB has led to various criticisms, particularly around weakened investigative management, less information for suspects regarding the progress of their case, and weaker protections for victims and witnesses.

The Home Office committed to reviewing the PCB and RUI process to consider whether further change will ensure that bail is being used where appropriate and to support the police in the timely progression of investigations. To help inform this work, this evidence review seeks to address the following issues: (1) what data exists on the use and nature of PCB; and (2) what impact have the 2017 reforms had on the use of PCB in England and Wales. The evidence review also explores the very limited evidence base on the effectiveness of PCB. It is based on a structured evidence-gathering exercise of published and forthcoming research literature, and data from a range of sources.

The use and nature of pre-charge bail

Limited published research exists on PCB before the reforms in 2017. Most studies have found that around one-third of suspects arrested were placed on PCB in the years before 2017. Other analysis suggests it might be closer to two-fifths. Studies have also found that substantial proportions of bailed suspects are ultimately not charged, with nearly half of cases in studies by both Hucklesby (2015) and Phillips and Brown (1998) resulting in no further action (NFA).

Hucklesby’s (2015) detailed two-force study found that while the ethnicity of bailed suspects broadly reflects the profile of those arrested, White European suspects were more likely to be charged than those from other ethnic groups. In contrast, bailed suspects from black and minority ethnic backgrounds were more likely to have their cases end in NFA.

There is very limited information on the frequency and type of conditions imposed through PCB, the rationale behind their imposition or the extent of breaches. Hucklesby (2015) found vastly different approaches towards imposing conditions in her two study forces prior to the reforms, with one force not using conditions at all, and the other applying conditions in two-thirds of PCB cases. ‘Banning’ conditions, i.e. keeping away from people and places, were those most frequently applied.

Based on Freedom of Information (FOI) responses from 29 forces in England and Wales, analysis by The Law Society covering the period before and after the change in legislation provides some evidence of the impact to the 2017 reforms. For a sub-set of 27 forces, in the 12 months following the reforms, the number of suspects on PCB dropped by around 85%, from approximately 190,000 to 29,000. Home Office data show that PCB numbers started to recover in the second year after the reforms. Figures for the same 27 forces show use of PCB increased by 106% to approximately 60,500 in the year ending March 2019, and a further 92% in the year ending March 2020 to approximately 115,800.^{[footnote 1]} However, these data suggest that PCB remains well below pre-reform levels.

It is generally believed that the modest recovery in PCB volumes in 2018/19 and 2019/20 is linked to concerns around the police’s response to the reforms raised initially by HM Inspectorate of Constabulary and Fire & Rescue Services (HMICFRS, 2018), and the police service’s response to the publication of the National Police Chiefs’ Council’s (NPCC) guidance on usage in 2019.

After the reforms and the creation of the presumption against PCB, individuals were increasingly RUI. The Law Society FOI analysis suggests that there was a clear shift from releasing individuals on PCB to RUI. In 2017/18, approximately 177,000 individuals were RUI in the 29 forces in England and Wales that responded to the FOI.

Post reform, a substantial proportion of individuals released on PCB for the first 28 days have their post release status changed to RUI thereafter. This adds to the complexity of measuring PCB and RUI volumes in the post-reform landscape.

Force-level variations

Force-level data indicate a wide variation in the use of PCB in the year before the reforms. Comparing PCB use against arrest volumes (28 forces), seven forces had a PCB to arrest ratio of 50% or above in 2016/17, while six forces recorded ratios below 20%.^{[footnote 2]}

In the year following the reforms – for the same sample of forces – eight forces had a PCB to arrest ratio of below 2%, while six forces had ratios above 10%. While almost all forces saw marked reductions in the proportion of arrestees who were subject to PCB in the wake of the reforms, some virtually abandoned using PCB altogether, while other forces maintained use at a moderate rate.

Impact of the decreased use of PCB and the increased use of RUI following the 2017 reforms

The large-scale reduction in the use of PCB, especially immediately after the reforms, has raised concerns about the impact on victims, witnesses, suspects and the wider criminal justice system (CJS) process. There were particular concerns that those suspected of serious offences – violent and sexual offences, and those involving repeat offenders – were RUI rather than granted PCB, even though they often met the pre-conditions for PCB.^{[footnote 3]} By releasing a suspect under investigation, and not on PCB, the police cannot impose conditions to manage the suspect, which in turn raised concerns about the impact on the protection of victims and witnesses.^{[footnote 4]} Concerns were also raised about how the move toward RUI affected suspects and the management of investigations.

Impact on victims and witnesses

Using bail to either directly change an offender’s behaviour towards a victim or to allow victims and witnesses to feel safer has been viewed as one of the central benefits of PCB. However, robust evaluative evidence on this effect is not available. Much of the ‘theory of change’ appears to be through imposition and policing of conditions attached, rather than through PCB itself. Recent research commissioned by HMICFRS and HM Crown Prosecution Service Inspectorate (HMCPSI) highlighted that bail conditions often made victims feel safer, particularly in offences where there is a close victim-offender relationship. However, Learmonth (2018) found that victims were often critical of specific conditions and their enforcement. In the post-reform landscape, HMICFRS/HMCPSI (2020) also found that the police were rarely considering the views of the victim when deciding on whether to release the suspect on PCB or RUI, and, for the former, what, if any, conditions to impose.

Others have pointed to a series of secondary benefits of bail conditions. For example, HMICFRS (2019) considered the impact of bail conditions in domestic abuse cases and suggested that bail conditions could help inform other agencies’ decision making, particularly in protecting victims at home. By contrast, the absence of conditions made it more difficult to evidence the need for emergency housing or a protection order (PO).

While this review did not identify any robust outcome evaluations on PCB, a more substantive evidence base exists on the effectiveness of POs. The PO research was initially examined as part of the review because POs share some features of PCB with conditions. However, ultimately, it was concluded that the differences in the way POs are administered were too great for the evidence to be easily translated across to PCB.

Analysis was undertaken to explore whether the sharp fall in PCB in the first year of the reforms was associated with a change in offences closed with the outcome ‘victim does not support the investigation – suspect identified’.

Five forces more than doubled their ‘victim does not support – suspect identified’ outcome numbers in the year to March 2018, more than in any other year ending March 2017 to 2020. In addition, four of these five forces recorded low bail to arrest ratios of less than 5%, suggesting the use of PCB post reform was at extremely low levels in these forces in the year after the reforms. However, this does not provide evidence of a causal link.

Impact on investigative management

A second concern was that creating the presumption against PCB removed a critical feature of effective investigative management. Research (Hillier and Kodz, 2012; Hucklesby, 2015) found that officers, pre-reform, viewed PCB as a useful policing tool and that re-bailing suspects was necessary to effectively manage an investigation. Having fixed bail dates served as prompts for the investigative process. Losing these prompts and IT systems not being adapted to monitor suspects after their release are thought to have contributed to poorer supervision of investigations and extended investigation times.

Concerns about the relationship between PCB reform and the timeliness of investigations have been identified as impacting on suspects, victims and witnesses. In the most extreme cases, delays are thought to have affected a small number of offences where investigations are constrained by time limits for prosecution, resulting in cases being dropped (HMICFRS/HMCPSI, 2020).

Data show long run increases in typical investigation durations since the year ending March 2015. The average (median) time taken to charge suspects has more than doubled from 14 days in the year ending March 2016 to 33 days in the year ending March 2020. However, it is hard to discern the precise impact of the shift to RUI from other factors. It is likely that the general increase in the investigative durations is due to a complex cocktail of factors including increasing levels of demand on the police, such as more reporting of complex crimes, and it being increasingly common to retrieve and examine digital evidence. The move to RUI may also have been a contributing factor by weakening the discipline of investigative management.

The impact on suspects

There are two main dimensions to the impact on suspects.

The large-scale shift to RUI has led to weaker communications between investigators and suspects. The reforms sought to reduce the long durations that suspects found themselves on PCB, principally under conditions. However, because of the shift to RUI, suspects find themselves without the guaranteed access to information about the progression of their case that the bail checkpoints previously created. This creates uncertainty around the case and can contribute to lengthy investigations, leaving suspects ‘in limbo’.

The lack of monitoring of RUI suspects by forces has potential issues around the handling of biometric samples. In his 2019 annual report, the Biometrics Commissioner noted that owing to the restrictive nature of police IT systems, some forces were unable to provide data about the number of suspects RUI, with knock-on impacts on the effective management of biometric samples (Wiles, 2020). This might have led to some forces unlawfully holding biometric data.

1. Introduction

The police in England and Wales can grant pre-charge bail (PCB), also known as police bail, to individuals arrested on suspicion of committing a criminal offence, but where there are no grounds to keep them in detention while the investigation is ongoing. At this stage in the investigative process, there is insufficient evidence to charge the individual for the offence. PCB is used to manage the arrested individual during the investigation while evidence is obtained in relation to the offence. A person granted bail is periodically required to re-attend a police station. To secure compliance with the requirements of bail, conditions may be attached, such as no contact with a victim or restrictions on entering certain areas.

Prior to the bail reforms of 2017, there were growing criticisms that some suspects were spending extended periods on bail before they were charged with an offence or the case was closed as ‘no further action’ (NFA). The Government legislated through the Policing and Crime Act 2017 (PCA) to address these concerns. Specifically, it introduced measures to limit PCB periods, seeking to rebalance the police use of bail in the interest of fairness. The PCA introduced a presumption against using PCB unless its application was considered both necessary and proportionate in all the circumstances. It also provided clear statutory timescales and processes for the initial imposition and extension of bail. PCB currently operates under this legislation.

Since the reforms came into force (April 2017), the use of PCB has fallen – dramatically in the first year – mirrored by an increasing number of individuals ‘released under investigation’ (RUI) (The Law Society, 2019). However, individuals released under investigation are not subject to conditions,^{[footnote 5]} and there are no time limits set for the suspect to return to the police station.

The marked decrease in using PCB and the increased use of RUI have led to a new set of criticisms. In general, these criticisms can be grouped into three areas:

Victims and witnesses

Suspects, who often met the pre-conditions (see footnote 3) for PCB under the 2017 amendments, including those suspected of domestic abuse and sexual offences, were nonetheless being placed on RUI. This meant that victims and witnesses in these cases were at risk of not being protected through conditions that police bail could provide.

Investigative management

Weakened investigative management due to ending the requirement under PCB of fixed points when officers had to engage with a suspect directly. This, it was argued, might affect the timeliness of case progression. Additional delays in already long case durations might further impact negatively on victims, witnesses and suspects.

Suspects

Suspects not released on bail can be left RUI for long periods of time with no or irregular updates on the progress of their case, adding further uncertainty to the process.

The Government committed to reviewing PCB to consider the need for further change to ensure that PCB is being used where appropriate and to support the police in the timely progression of investigations.

As part of the review, it was proposed that an evidence review on PCB should be undertaken. Specifically, this has sought to address two key issues: (1) what data exists on the use and nature of PCB; and (2) what impact have the 2017 reforms had on the use of PCB in England and Wales. The exercise also intended to explore the evidence on the effectiveness of PCB. However, very limited robust evidence directly associated with the effectiveness of PCB was identified.

2. Methodology

The contents of this review are based on a structured evidence-gathering exercise of published evidence and data.

2.1 Research evidence

The research literature assessed for this review were studies which were easily accessible and in the English language. The search for international research on PCB focused on criminal justice processes which were equivalent to PCB in England and Wales. Within the UK, broadly similar approaches to PCB were found in Scotland and Northern Ireland, although they have different legislative terms. For example, in Scotland the process for releasing suspects before charge with conditions is called ‘investigative liberation’. Hence the search criteria needed to be sufficiently broad as to identify the legislative and criminal justice processes in other countries.

An initial exploratory review, and discussion with subject matter experts, revealed that, while a reasonable amount of research and grey literature exists on the nature and use of PCB, there was little published research into the effectiveness of PCB. However, a more extensive evaluation-based research literature did exist on other forms of pre-charge ‘controls’ such as protection orders (POs). PO interventions, while not direct equivalents to PCB, bore some similarities to PCB with conditions in the way they sought to address offender behaviour, or to support victims, before any charges against a suspect were laid. POs were therefore initially included within the search criteria.

The full set of search criteria used were:

Pre-charge bail

Police, law enforcement, court, pre-charge bail, bail, released under investigation, post-charge bail, crime, policing, criminal justice, domestic abuse, effectiveness, domestic violence, sexual assault, harassment.

Protection orders

Protection orders, civil protection orders, domestic violence protection orders and notices, barring orders, civil orders, stalking protection orders, restraining orders, effectiveness, occupation orders, non-molestation orders, domestic abuse.

Search terms were applied to the following databases: National Police Library, Social Research Association, Campbell Collaboration, NCJRS, WorldCat, Google Scholar and Google search. This search was limited to studies from 1984 to present, to only include research on PCB following legislative changes contained within the PACE Act (1984). The main searching took place between November 2019 and March 2020.

The search yielded 79 published articles and reports which initially met the search criteria for PCB. Three studies were also provided by academics with expertise in PCB. Of the articles and reports, many were excluded because they were not empirical research studies or academic evaluations, or they were not directly relevant to the topic area. This report draws primarily upon eight studies which involved exploratory research PCB or RUI and 11 studies which explored bail in the wider context of the criminal justice system. None of these studies were formal impact evaluations and the majority were descriptive. None reported direct evidence to answer the question on the effectiveness of PCB within the criminal justice process.

The search yielded 61 published studies which initially met the search criteria for POs. Of these, 13 were subject to a more detailed review. There were three systematic reviews and/or meta-analyses (Benitez et al., 2010; Dowling et al., 2018a; Cordier et al., 2019). One focused specifically on reviewing the effectiveness of POs in reducing recidivism in domestic violence^{[footnote 6]} cases (Cordier et al., 2019). The other looked more broadly at research into the use and impact of POs for domestic violence, using the EMMIE framework^{[footnote 7]} (Dowling et al., 2018a). The literature covered by both reviews is dominated by research on POs from the US, therefore it may not be generalisable to England and Wales. The third review, by Benitez et al. (2010), explored whether POs protect victims and the public by reducing the risk of future harm. The ten remaining individual PO studies were reviewed using the Maryland scale.^{[footnote 8]}

Ultimately, having reviewed the PO evidence, it was concluded that the differences in the way POs are administered were too great for the evidence to be easily applied to PCB. The findings from the review of studies of POs are therefore summarised only briefly in the main report, with a more extensive description in Appendix D.

2.2 Data

The evidence review has also drawn upon available data on the use of PCB in England and Wales. These were frequently patchy and inconsistent (HMICFRS/HMCPSI, 2020). Data on the use of PCB were not routinely collected until after the reforms in 2017. Beyond the bespoke studies described above, very little data have been collected on bail conditions, bail breaches, offending on bail, bail extensions and police actions in response to any bail breaches.

Freedom of Information (FOI) requests generated some useful material on overall use of PCB, although there were issues with both coverage – not all forces responded – and comparability across forces (The Law Society, 2019). Differences exist in recording and categorising bail from force to force (e.g. individual force IT systems for bail management are not analogous) making it difficult to precisely compare PCB usage across forces. HMICFRS/HMCPSI (2020) identified several reasons why the data may not be comparable. Forces measure PCB and RUI cases in different ways. For example, some forces count on an offender basis, whereas others count on an offence basis. Nevertheless, the FOI data were helpful as they allow, to some extent, broad comparisons to be made on the use of PCB before and after the reforms.

Data on PCB have been published as experimental statistics in the Home Office’s Police Powers and Procedures statistical bulletins since 2018 (Home Office, 2018; 2019; 2020a). These cover how many individuals were released on PCB and average time spent on PCB. In the first year, year ending March 2018, coverage was limited to only 17 forces, but this increased to 41 forces and 40 forces in years ending March 2019 and 2020, respectively. Despite data now being routinely collected, police forces^{[footnote 9]} and HMICFRS have raised concerns about the quality of data, highlighting the existence of partial returns and a lack of consistency between police forces in recording PCB (Home Office, 2019; HMICFRS/HMCPSI, 2020). As these data are experimental, any conclusions should be made with caution. The experimental statistics do not cover RUI data.

3. History and context of pre-charge bail

There are several stages in the criminal justice process when the decision to bail an individual can be made:

• By the police when there is not sufficient evidence to charge, also known as precharge bail (PCB).
• By the police when there is sufficient evidence available to charge, also known as post-charge bail.
• By the courts when a person has been charged, also known as court bail.

Bail can be imposed with or without conditions in all three circumstances.

3.1 Process of pre-charge bail

The police in England and Wales can grant PCB under Part 4 of the Police and Criminal and Evidence Act 1984 (PACE). The suspect can only be detained in custody in most cases up to 24 or 36 hours (referred to as the ‘custody clock’) before they must be charged or released. If additional enquires are required, and they cannot be undertaken immediately, then the suspect is released until the police complete their enquiries, so they are not detained in custody unnecessarily. PCB means the individual under investigation is released from police custody, with or without conditions, while officers continue their investigation.

In the period before April 2017, an individual who was arrested, but for whom enquiries were ongoing, would usually be released on PCB. On rare occasions, releasing a suspect pending further investigations without bail prior to the 2017 reforms took place in the aftermath of Hookway.^{[footnote 10]} Since April 2017, suspects released without charge are either RUI or on PCB, with or without conditions.

Once released from custody on PCB, the suspect’s custody clock stops and when the suspect returns to answer their bail, the clock continues and any further evidence that has been obtained can be put to them. The suspect may then be charged, re-bailed, released under investigation or no further action taken against them. The Bail Act 1976 specifies that no conditions should be imposed with PCB unless it appears to the officer granting bail that it is necessary to do so to prevent a person from failing to surrender or committing further offences etc (see 3.2 below for more details) (Bail Act, 1976). Legislation covering bail exists under multiple Acts and Figure 1 sets out how the authority for the police to impose PCB has evolved over time.

Figure 1: Timeline of PCB in England and Wales

Year	Event
1925	Pre-charge bail introduced under the Criminal Justice Act (CJA) 1925
1976	Following the CJA 1967 which introduced conditional bail, Section 3 of the Bail Act 1976 enshrined the power to impose conditions on bail for courts
1984	Following the 1981 Royal Commission on Criminal Procedure, PCB subjected to tighter controls under The Police and Criminal Evidence Act (PACE) 1984
2003	Police powers extended to allow officers to attach bail conditions to PCB under the CJA 2003
2017	Reforms to PCB following the consultations were introduced in the PCA 2017
2019	Review into PCB law announced by Government

Source: Bail Act (1976); PACE (1984); CJA (2003); Cape (2016); PCA (2017); Gov.uk (2019)

Information on comparative jurisdictions, such as Northern Ireland and Scotland, is summarised in Appendix A. PCB can also be imposed where there is sufficient evidence to charge a suspect and the suspect is bailed while the CPS make a charging decision (granted under s.37(7)(a) PACE). This type of pre-charge bail was not the subject of the reforms made under the PCA 2017 and is not subject to the same authorisations or timescales as PCB where there is insufficient evidence to charge. It is not covered in detail in this evidence review.

3.2 Pre-charge bail with conditions

The Criminal Justice Act (CJA) 1967 introduced conditional bail which was enshrined in the Bail Act 1976 and subsequent legislation has amended it further.^{[footnote 11]} If the preconditions (see footnote 3) for granting bail as set out in Section 3 of the Act are met, conditions should be applied to the suspect’s bail in order to ensure that the suspect:

• surrenders to custody at the end of the bail period
• does not commit an offence while on bail
• does not interfere with witnesses
• does not otherwise obstruct the course of justice.

Conditions may also be attached for the suspect’s own protection and, where the suspect is a child or young person, for their own welfare. Conditions imposed may include:

• A ban on leaving the country including the requirement to surrender a passport
• Not being allowed to enter a certain area such as the home of the alleged victim
• Adherence to curfew
• Residing at a specific residence during the bail period
• Not being allowed to communicate with certain people e.g. victims, witnesses and known associates.

Conditions are not legally specified and central advice or guidance on what conditions police should impose is limited. There are very few restrictions on what conditions they can impose, with officers using their own discretion. However, the Bail Act 1976 and the CJA 2003 state that for either pre-or post-charge bail the police cannot impose the following conditions on a suspect (CPS.gov.uk, 2019):

• To reside at a bail hostel
• To attend an interview with a legal adviser
• To make themself available for probation enquires and reports
• That contain electronic monitoring requirements.

Taking this all in to account, PCB can arguably be claimed to be meeting three main objectives: suspect management (including preventing re-offending); protecting victims and witnesses, primarily through conditional PCB; and investigative management including allowing time to obtain evidence.

Purposes of unconditional PCB

• Investigative management, e.g. allows time to obtain evidence, investigations to be conducted in a timely manner and limits the time spent in police custody
• Suspect management, which is closely linked to investigative management, e.g. suspect under duty to re-attend a police station

Purposes of conditional PCB

• Protect victims and witnesses, e.g. to ensure suspect does not approach or contact the victim or witness
• Investigative management, e.g. does not obstruct the course of justice
• Suspect management, which is closely linked to investigative management, e.g. preventing further offences being committed while on bail or keeping suspects away from criminal associates

3.3 Breaches

Under PCA 2017, breach of PCB conditions is arrestable, but it is not a criminal offence (CPS.gov.uk, 2019). Therefore, there is no criminal sanction resulting from a breach.^{[footnote 12]} Officers can arrest individuals for a breach, and then charge the suspect with the original offence or release them with or without charge, either on bail or without bail. If they are released on bail, conditions set for the original bail can be re-applied. There is currently no available data on breaches of conditions, although some exploratory research on this area (Hillier and Kodz, 2012) is discussed later in this report.

3.4 The 2017 Act and its aftermath

Calls for legislative reforms to PCB grew following several high-profile cases where individuals were released on bail for long periods – in some cases over a year – but ultimately not charged. These cases included individuals arrested on suspicion of sexual offences between 2012 and 2014 as part of Operation Yewtree. Bail conditions included having supervised contact with their children and surrendering passports. When giving evidence to the Home Affairs Select Committee in March 2015, one celebrity who was arrested and bailed without charge expressed the view that the protracted period on bail resulted in significant personal, professional and financial loss (House of Commons Home Affairs Committee, 2015).

Following public consultations by the College of Policing (2014) and the Home Office (2015), and on the back of the Home Affairs Committee report (2015), further legislation regarding PCB was introduced. The Policing and Crime Act (2017) came into force on 3 April 2017 which aimed to rebalance the police use of bail in the interest of fairness. This legislation included reforms of PACE (1984) such that:

• an initial PCB limit of up to 28 days was set
• a presumption of release without bail unless necessity and proportionality criteria are met
• extensions of PCB had to be authorised by a superintendent (up to three months after the initial bail date), or by a Magistrates’ Court (longer than three months after the initial bail date)
• the authorising inspector must consider any representations made by the person or their legal representative (s.50A(b) PACE).

Historically, the purpose of PCB was to limit the time spent in police detention and to release suspects, without charge, for them to return later to the police station. Under PACE it was outlined that PCB is available to manage an arrested individual while the investigation continues to obtain evidence in relation to the offence, or while a decision to charge is being attained.

However, the PCA 2017 introduced further bail pre-conditions which had to be met. The custody officer now needs to feel satisfied that releasing the person on bail is necessary and proportionate in all circumstances. In addition, an officer of the rank of inspector or above must authorise the release. If this is not the case but the grounds remained to suspect an individual of an offence, then they should be released with no bail obligations (i.e., RUI).

When individuals are RUI, they are not subject to any conditions, and as there are no time limits set for the suspect to return to the police station, the time for the investigation is unlimited. This is contrary to the intentions of the amendments of the PCA (2017), which aimed to reduce how long an individual was under investigation.

There was a substantial reduction in PCB in the immediate aftermath of these changes, mirrored by an increase in using RUI. The marked decrease in using PCB, and the growth in RUI, has led to a new set of criticisms. The Law Society (2019) drew on case studies from lawyers where suspects had been RUI for long periods with no updates on how their case was progressing. The Centre for Women’s Justice (2019) launched a super-complaint which heavily criticised the police for failing to impose bail conditions in high-risk cases where female victims were particularly vulnerable. There were also concerns that the reforms had led to weakened investigative management, particularly with the increased use of RUI, as there were no requirements for officers to check in with suspects.

HMICFRS’s PEEL effectiveness report (2018) for 2017 highlighted concerns about the reduced use of PCB. It included a national recommendation for all forces to review how they implemented the PCB reforms by September 2018. Subsequently, the National Police Chiefs’ Council (NPCC) published guidance around using PCB and RUI in early 2019. This guidance was intended “to reinforce the message that in the right circumstances, the use of PCB is still a legitimate investigative and safeguarding tool” (NPCC, 2019; p. 1). The Government announced in 2019 that it would review PCB laws (Gov.uk, 2019.)

4. The use and nature of pre-charge bail

4.1 The use and nature of PCB prior to the 2017 reforms

Very little published data exists on the use and nature of PCB. One rare early insight comes from Phillips and Brown’s (1998) detailed statistical examination of the CJS process in the 1990s. This was based on a survey of 4,250 people who had been detained at ten police stations in England and Wales. Research by Hillier and Kodz (2012) and Hucklesby (2015) provides useful insights into the use of PCB prior to the reforms. Hillier and Kodz (2012) conducted interviews and focus groups with 80 police practitioners in five police forces. Hucklesby (2015) conducted research in two police forces in England between 2011 and 2013. The mixed methods study comprised observations, a review of 14,173 PCB records, questionnaires completed by police officers and 38 interviews with police staff. Martin (forthcoming) undertook a detailed study of PCB two years before and two years after the reforms in a single force. His dataset included over 247,000 individual offences across the four-year period.

Volumes

Several studies suggest that, pre-reform, approximately one-third of those arrested were subsequently released on PCB (see Appendix B; Hillier and Kodz, 2012; House of Commons Home Affairs Committee, 2015; College of Policing, 2016; College of Policing, 2017). Other analysis has suggested a higher proportion. Based on data from 12 police forces provided by the College of Policing, it has been estimated that 404,000 individuals were released on PCB in the year ending March 2014 (Home Office, 2015), equivalent to around 43% of arrestees. And in the year immediately before the reforms (2016 to 2017), a comparison of FOI data and Home Office arrest data suggested a similar overall ratio (Law Society 2019; Home Office 2020a).^{[footnote 13]} However, these aggregate estimates conceal much wider variations in PCB use at force level.^{[footnote 14]}

Offence type

Hucklesby (2015) found that PCB was most commonly used for: violence offences, approximately one-third in both forces; theft-related offences, around one-fifth in both forces; and property offences (19% in Force A, 13% in Force B). Philips and Brown (1998) also found that the use of PCB varied considerably according to offence type. It was most frequently given where suspects had been arrested for fraud, sexual offences, robbery and drug offences, where approximately 30% of those arrested were bailed for further enquires.

Duration

Based on data provided by 12 forces for the year ending March 2014, around eight in ten (79%) suspects were on PCB for up to three months, while an estimated 14% were on bail for between three and six months (Home Office, 2015). Only 1% – equivalent to an estimated 5,000 individuals nationally – were on bail for in excess of 12 months. Hucklesby (2015) found that the average time spent on bail was between six and seven weeks in both forces studied, in line with a College of Policing (2016) estimate based on nine forces. The College of Policing (2016) also found that of the 9% of cases bailed initially for over 90 days, rape, sex and drug offences were most common (accounting for 55% of these cases). As with most aspects of PCB, individual forces show wide variations in the average duration of PCB. Law Society (2019) data indicate that in the year prior to the reforms, the average duration of PCB for suspects in nine forces ranged from 53 days to 175 days. Martin’s (forthcoming) single force study, based on data for suspects linked to almost 25,500 offences recorded between April 2015 and April 2017 (i.e. before the reforms), found an average PCB duration of 140 days per offence, with 65% of offences dealt with within the first three months.^{[footnote 15]}

Outcomes post bail

Studies have generally found that around a half of PCB cases before the reforms resulted in NFA. Hucklesby (2015) found that almost half of the bail records analysed in the two police forces had an NFA outcome while Phillips and Brown’s (1998) much earlier study found that 44% of PCB cases had an NFA outcome. Hillier and Kodz (2012) found that those bailed were more likely to be given an NFA outcome than a charge. The proportion of NFAs has also been found to vary by offence type, with 60% for sexual offences, 50% for violence offences and 20% for traffic offences (Hucklesby, 2015).

Ethnicity

In her research on two police forces, Hucklesby (2015) found that while the ethnicity of bailed suspects broadly reflected the profile of arrest data, White European suspects given PCB were more likely to be charged than people from other ethnic groups on PCB. Cases involving suspects from black and other minority ethnic backgrounds were more likely to result in NFA.

Conditions

There is little data on the frequency and type of conditions imposed through PCB or the extent to which suspects comply. This extends to data on reasons behind decisions and breaches of conditions. However, the evidence that exists suggests that practice can vary widely from force to force. Hucklesby (2015) found that of the two forces, one used bail conditions in 67% of PCB cases. The other force had a policy not to impose bail conditions on PCB at all, which was criticised by nearly all those interviewed in the force. Martin’s (forthcoming) single force study found 68% of suspects on bail had conditions in the year before the reforms. Hucklesby’s (2015) interview-based research suggested that ‘banning’ conditions, i.e. keeping away from people and places, were the type of conditions most frequently used. Police officers interviewed by Hillier and Kodz (2012) viewed penalties for breaching police bail conditions as inadequate, commonly describing them as being a ‘toothless tiger’.

There is a larger evidence base on conditions around post-charge and court bail; both Hucklesby (2001) and Raine and Wilson (1996) found regional variation on the use of post-charge bail conditions in their research. The types of conditions focused around a few themes such as residing at a specific address and similar to Hucklesby’s (2015) findings for PCB, ‘banning’ conditions’ include preventing contact with victims and/or witnesses, and keeping away from specific areas and associates.

4.2 The use of PCB and RUI in England & Wales following the reforms in 2017

The number and percentage of individuals on PCB

Arguably the most useful source of data on the immediate impact of the 2017 reforms on the use of PCB comes from a Law Society (2019) report. Their analysis examined the use of PCB and RUI by the police in England and Wales and Northern Ireland. The main value of The Law Society analysis is that it covers the period before and after the change in legislation. However, these data are not part of an established data collection but are based on an FOI request and should therefore be treated with caution.

There may also be some specific measurement issues in PCB data in the post-reform period. Pre-reform, suspects would stay on PCB for the duration of the investigation. However, this situation seems to have changed in the post-reform period. Martin’s (forthcoming) detailed study of an English force found that while individuals would be placed on PCB initially, some switched to RUI after the initial 28-day bail period. This affected 20% of bailed cases in the year ending 2 April 2019. Of the offence types analysed, sexual offences (26%), rape/attempted rape (19%) and grievous bodily harm (18%) were most likely to switch from PCB to RUI following the initial 28-day bail period.^{[footnote 16]} Home Office exploratory research^{[footnote 17]} undertaken for this review also found evidence that, post-reform, PCB cases were often switching to RUI after the initial 28-day bail period (between 31% and 37% of cases in the two forces examined). This makes the question of consistently counting PCB and RUI volumes much less straightforward after the reforms. The HMICFRS/HMCPSI thematic review of PCB (2020) found that, in most forces, a suspect’s status defaults to RUI after the 28-day initial bail period ends.

The Law Society (2019) report is based on data from an FOI request submitted by law firm Hickman & Rose Solicitors. In response to the request, 29 police forces in England and Wales returned data for years ending 2 April 2017 and 2018.^{[footnote 18]} The following analysis combines FOI data from The Law Society and Home Office experimental statistics^{[footnote 19]} for the years ending March 2019 and 2020, to provide an overview of trends based on a consistent sample of forces.^{[footnote 20]} In Figure 2, a subsample of data from 27 forces for all datasets is presented to map aggregate trends in PCB use. This subset of 27 forces account for three quarters (75%) of PCB volumes in the year ending March 2020.

Figure 2: Number of individuals arrested and subsequently placed on PCB in England and Wales, based on a sample of 27 forces (Based on a consistent set of 27 English and Welsh forces covered in all three datasets.)

Year	Number placed on PCB
2016/17	190000
2017/18	29300
2018/19	60500
2019/20	115800

Figure 2 shows that, in 27 forces, approximately 190,000 individuals were put on PCB in year ending 2 April 2017 (Law Society, 2019). This figure dropped by 85% to approximately 29,300 in the year ending 2 April 2018, the year after the reforms. Home Office experimental statistics for the same 27 forces suggest the number of individuals on PCB increased by 106% in the year ending March 2019, to 60,500. In the year ending March 2020, numbers increased again, up by 92%, to 115,800, but were still below prereform volumes (a difference of around 74,000). The Home Office experimental statistics on PCB achieved high levels of force coverage in the most recent two years (41 forces in the year ending March 2019, and 40 forces in the year ending March 2020). Looking at the 40 forces which submitted data for each of the last two years, there was an 81% increase in the use of PCB, from approximately 84,200 to 153,500, in England and Wales (Home Office, 2019; Home Office, 2020a).

The increase in PCB after the reforms were introduced in April 2017 is thought to be partly related to concerns raised by HMICFRS and the issuing of NPCC guidance in 2019. In their PEEL effectiveness report for 2017, HMICFRS issued a recommendation requiring all forces to review how they implemented changes to PCB by September 2018. That report also stated that “forces should then put into effect any necessary changes to make sure they are using bail effectively, and in particular that vulnerable victims get the protection that bail conditions can give them” (HMICFRS, 2018; p. 29). NPCC operational guidance clarifying the position around the use of PCB was published in January 2019 (NPCC, 2019). This guidance emphasised the need to continue to use PCB, where necessary and proportionate, highlighting PCB’s importance as a legitimate investigative and safeguarding tool.

While issues remain about the consistency and reliability of the FOI data across forces, it is helpful to examine the broad pattern of PCB usage by force, before and after the reforms. Force by force numbers are given in Appendix C. Figure 3 sets out ratios of PCB to arrest volumes for 2016/17 and 2017/18 ranked by 2018 values.^{[footnote 21]} They imply a very wide variation in use of PCB in the year before the reforms. Seven forces achieved ratios of 50% and above in 2016/17, while six forces recorded ratios of below 20%. Following the reforms, eight forces saw ratios fall to 2% or less. However, six forces^{[footnote 22]} had ratios of 10% and above, with Essex having the highest ratio (24%). Although these ratios can only be taken as illustrative, they suggest that while most forces saw marked reductions in the PCB to arrest ratio across the two years, some saw the ratio fall to very low levels. And while all but one force reduced their ratios, others achieved ratios which exceeded those that some forces were achieving before the reforms.

Figure 3: Estimated ratio of PCB volumes to arrests, years 2016/17 and 2017/18, selected forces

There is a two-day difference between the timeframes for the arrest data collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by The Law Society (3 April to 2 April).

Force	2016/17	2017/18
Hertfordshire	8.8	0.6
Cleveland	14.0	0.7
Dorset	37.4	1.2
Thames Valley	39.8	1.2
Derbyshire	5.1	1.5
Devon and Cornwall	40	1.8
Leicestershire	44.8	2.2
Avon and Somerset	41.3	2.3
North Wales	33.4	3.7
Staffordshire	38.5	3.7
Northamptonshire	33.8	4.1
Nottinghamshire	50.9	4.2
Cheshire	35.9	4.7
Lincolnshire	38.2	5
Merseyside	21.6	5.5
Gwent	50.2	6.9
Bedfordshire	40	6.9
Greater Manchester	39.6	6.9
London Metropolitan	44.6	7.9
Cumbria	51.1	8.3
Cambridgeshire	48.45	8.7
Norfolk	4.5	8.8
Warwickshire	50	10.7
Surrey	57.3	11.1
Suffolk	12.5	12.2
South Wales	21.6	13.7
West Mercia	55.2	14.8
Essex	71.	24.2

Source: Law Society (2019); Home Office (2020a)

Limited data exist on how the PCB reforms affected different crime types. Martin’s (forthcoming) detailed before and after study of one force found the overall composition of PCB by offence type changed after the reforms. Comparing the percentage share of PCB by offence type, Martin found there was a shift towards more serious interpersonal offences in the two years after the reforms, and a corresponding fall in the share accounted for by theft offences. For example, sexual offences increased from 10% to 18% of total PCB and GBH went up from 6% to 12%. By contrast, theft offences fell as a share of PCB (27% pre-reforms but only 8% in the two years after). However, changes in the crime type share of PCB need to be considered against the marked fall in its overall use after the reforms. Martin found that total volumes fell from an average of 12,344 prereform (April 2015 to April 2017) to an average of 1,165 post-reform (April 2017 to April 2019). So, while serious offences accounted for a greater share of all PCB after the reforms, bail volumes for these offences showed absolute falls in this same time period.

Use of RUI following the reforms

After the reforms, and introducing the presumption against PCB, the main option for managing arrested suspects whose investigations were ongoing became RUI. Just as there had been a large fall in the use of PCB in the immediate aftermath of the reforms, the use of RUI became commonplace (The Law Society, 2019). Figures based on FOI requests from 29 forces in England and Wales published by The Law Society revealed that approximately 177,000 individuals were on RUI in the year ending 2 April 2018 (The Law Society, 2019). This broadly equates to the approximate 171,000 reduction in the number of arrestees on PCB for these forces, suggesting that there was a simple shift from releasing individuals on PCB to RUI. However, we would not expect these figures to match entirely. One of the challenges around measuring PCB (and RUI) volumes post-reform is that the individuals would sometimes shift from PCB to RUI as the case progressed (see above).

Timeliness

Comparing data on durations for those on bail is arguably of only limited value. This is because the precipitous reduction in bail volumes that took place after the reforms mean that neither the volume of cases, nor the types of offence characteristics covered, are comparable before and after 2017.

5. The impact of the decreased use of PCB and increased use of RUI following the reforms in 2017

The large-scale reduction in using PCB, especially in the period immediately after the reforms, raised concerns about the impact on suspects, victims, witnesses and the wider criminal justice system. In particular, there were concerns that those suspected of serious offences – violent and sexual offences – and those involving repeat offenders, were being RUI rather than released on PCB, even though they often meet the pre-conditions for PCB (see footnote 3) (see for instance, The Law Society, 2019). In this section, evidence on the impact of the changes is considered.

5.1 Views of the impact on victims and witnesses – primary and secondary effects

One area of concern is how reduced use of PCB has affected victims’ and witnesses’ safety and wellbeing, especially in cases involving domestic abuse or other offences involving close victim-offender relationships, for example rape, sexual assault and stalking offences (Learmonth, 2018). There are several linked questions here. The first of these – classified here as primary effects – relates to how the use of bail conditions can either change an offender’s behaviour towards a victim, or, alternatively, simply allow a victim to feel safer (Centre for Women’s Justice, 2019).

5.2 Victim perceptions on pre-charge bail

Studies which have focused explicitly on the extent to which victims felt bail affected their feelings of safety are rare. As part of the HMICFRS/HMCPSI inspection on PCB, research was conducted with 27 victims of crime by BritainThinks (2020). Participants had been victims of a range of crime types, but domestic abuse and stalking/harassment offences dominated (17 of the 27 participants). Investigations had been closed at the time of interview, with offences taking place within the 18 months before their participation in the research. Some victims were not clear whether suspects in their cases had been given RUI or PCB, and generally found the terminology confusing. However, participants were apparently clear about when suspects in their cases had been released under PCB with conditions. Participants generally described feeling safer in cases where suspects had been released with bail conditions, particularly if they knew that the suspect was not permitted to contact them while the investigation was ongoing. Knowing that they had ‘permission’ to contact the police and report this if it happened made them feel more confident, even if they believed the likelihood of a breach of this condition was high. These findings echo those from some of the literature on POs – see Appendix D. By contrast, some victim interviewees described feeling unsafe when suspects were under RUI. They felt there was nothing in place to prevent re-offending, a view most strongly held by domestic abuse and stalking and harassment victims. An exploratory small-scale study into the use of bail in rape cases (Learmonth, 2018) found that survivors expected to be entitled to personal safety during the investigation and were distressed when bail conditions were felt to be inadequate.

5.3 Professionals’ perceptions of victims’ security

Research undertaken for HMICFRS by BritainThinks (2020) also included interviews with professionals – mainly those working in third-sector bodies directly supporting victims – to gather their perceptions on the impact of PCB on victims’ feelings of safety. Two main findings emerged. First, they argued that the absence of bail conditions signalled to victims that even if they are at risk they were being taken less seriously, and they felt ‘isolated’ and ‘alone’. Second, they argued that absence of conditions resulted in more limited lines of communication with the police. This in turn engendered a stronger sense of support from the police through PCB (with conditions) than RUI.

There is certainly some evidence that, three years after the reforms, and despite the recovery in PCB volumes, the right balance between bail and RUI has not been found. The joint HMICFRS/HMCPSI inspection on PCB (2020) was critical of the police for not considering the views of the victim when deciding on whether to bail a suspect and informing the conditions to impose. In their analysis of 140 case files in charged cases, they found that in just under half (62), RUI had been used when PCB with conditions would have met the ‘necessity’ and ‘proportionate’ tests. These offences included domestic abuse, sexual offences and offences against children. Martin’s (forthcoming) research in one force found that for offences which had warning classification flags for ‘domestic abuse’ and/or ‘child at risk’, the proportion of suspects who were subject to PCB fell markedly when comparing the two years before and two years after the reforms (from 15% down to 3%). The Law Society (2019) also highlighted several examples where not putting the suspect on bail put the public at risk.

This, of course, is not the same as saying the system was working effectively in managing the risk around victims before the reforms, particularly in domestic abuse and sexual offence cases. The evidence points to a variable picture, albeit one where a much higher number of suspects were placed on PCB for the duration of their investigations prior to the reforms. But in the area around managing victim risk, on the most critical issue around the use conditions, very limited evidence exists. Hucklesby’s (2015) research on two forces showed that one force operated with no conditions, while the other force used conditions in two-thirds of PCB cases. Learmonth’s (2018) small-scale study found four of the six women interviewed to be critical of the conditions of PCB which were insufficient for their needs, particularly the lack of consequences for breaches of conditions.

5.4 The evaluation evidence base

Of course, we still cannot draw on robust evidence that shows that PCB with conditions definitively works, leading to better outcomes (e.g. reduced victimisation, higher levels of victim satisfaction or lower levels of victim withdrawals, reduced risks to the suspect). No evaluative studies were identified explicitly on PCB. Some potentially useful evaluations conducted on the effectiveness of POs were identified and are summarised in Appendix D. Some functions of POs are similar to conditional PCB, such as applying restrictions on individuals not to contact the victim or enter certain locations or addresses. However, there are some fundamental differences between PCB and POs that make ‘translation’ of the evidence difficult. In some jurisdictions, PO violations can result in a criminal sanction, which is not the case for breaches of PCB. Durations of POs are typically longer than PCB. POs tend to be victim-generated rather than an automatic criminal justice response initiated by the police and tend to be focused on a smaller subset of offence types (principally around domestic abuse).

The evidence base on the effectiveness of POs suggests that there are some situations where POs may be effective in reducing re-offending and protecting the victim. Two systematic reviews provided some evidence that POs can be effective at reducing recidivism in domestic violence and interpersonal violence cases (Dowling et al., 2018a; Cordier et al., 2019). In their evaluation of Domestic Violence Protection Orders (DVPOs) in England and Wales, Kelly et al. (2013) found that DVPOs were associated with reduced rates of victimisation. However, other studies found that some victims continued to suffer violence or violations of their POs (Logan and Walker, 2009; Kothari et al., 2012). Nevertheless, Logan and Walker (2009) found that most victims perceived POs to be effective, even in cases where POs were violated and violence continued. Kelly et al. (2013) also found that POs could be more effective when used alongside support services. But, as noted, directly translating these findings across to PCB with conditions is not straightforward.

5.5 Changes in ‘victim does not support – suspect identified’ outcomes

Data were examined to assess whether any impact of reduced use of PCB might be reflected by an increase in offences closed with the police outcome ‘evidential difficulties suspect identified – victim does not support further action’.^{[footnote 23]} The assumption was that victims might be less likely to support investigations if suspects in their cases were not subject to PCB (with conditions), and any effect might be most apparent in those cases where victims and suspects were known to each other. The precise theory of change is likely to be complex, but the absence of bail with conditions might be linked to an increase in victim’s concerns around safety or a feeling that a case is not being taken seriously and increase the likelihood of victim withdrawal. In addition, if the ending of PCB contributed to longer investigations, this also might increase the risk of victim withdrawal. If there was a year-on-year impact on ‘victim does not support’ outcomes in the wake of the reforms, it would be most evident in the year ending March 2018, as the partial recovery in PCB volumes in subsequent years might have ameliorated any immediate effect.

Many factors lead victims to withdraw their support for police action. An international review of the evidence on domestic abuse – ‘victim does not support’ outcomes are overrepresented within DA offences^{[footnote 24]} -found a variety of reasons for victims’ retraction of statements (Dowling et al, 2018b).^{[footnote 25]} The most commonly identified were: fear of reprisal; still wanting a relationship with the perpetrator; wanting the perpetrator to receive help instead of punishment; not wanting their children to be without a parent; not wanting to subject their children to the court process; fatigue with or pessimism regarding the court process; and financial reliance on the perpetrator. A casefile-based study of recorded rapes (Feist et al., 2007) found that, in victim withdrawal cases, victims wanting ‘to move on’ and a reluctance to see through the investigation or prosecution each accounted for one-fifth of withdrawals. Pressure from others to withdraw and fear of reprisals accounted for 10% and 5% respectively. In addition, other factors will affect changes in ‘victim does not support’ volumes over time. These are likely to include changes arising from changes in the reporting of offences, the composition and recording of offences and the recording of outcomes.

Table 1 and Table 2 show trend data for ‘victim does not support’ outcomes – suspect identified’ with ‘suspect not identified’ included for comparison. The number of ‘victim does not support – suspect identified’ outcomes grew steadily in the period after year ending March 2015, increasing by more than fivefold by the year ending March 2020, accounting for a steadily increasing share of rising recorded crime volumes (Table 1 and Table 2). At the aggregate level, it is hard to find evidence of a sudden and marked shift in ‘victim does not support – suspect identified’ outcomes taking place shortly after the reforms. The annual rate of increase slows down during the period, including between the critical years ending March 2017 and 2018 (down from 38% to 34% respectively). We have also looked at changes in the ‘victim does not support – suspect identified’ outcome by force, owing to the variability in the use of PCB at the force level before and after the reforms.

Table 1: Trends in the number of outcomes(a) recorded in year, by selected outcome type:years ending March 2015 to 2020 (excluding GMP^{[footnote 26]})

The figures in this table relate to outcomes recorded in the year regardless of when the associated crime was recorded.

Years ending 31 March	2015	2016	2017	2018	2019	2020
Evidential difficulties, suspect identified: victim does not support further action	194,036	392,679	542,671	726,235	906,054	1,009,432
Year-on-year percentage change		102%	38%	34%	25%	11%
Evidential difficulties, suspect not identified: victim does not support further action	49,443	118,618	158,931	216,066	250,609	239,062
Year-on-year percentage change		140%	34%	36%	16%	-5%

Source: Home Office (2020b)

Table 2: Trends in the ratio of selected outcomes recorded in year to total recorded crime, years ending March 2015 to 2020 (excluding GMP) (The figures in this table relate to outcomes recorded in the year regardless of when the associated crime was recorded. Ratio based on number of outcomes recorded in year divided by number of crimes recorded in year.)

Years ending 31 March	2015	2016	2017	2018	2019	2020
Police recorded crime	3,374,272	3,671,883	4,056,404	4,551,859	4,935,419	5,006,941
Change		9%	10%	12%	8%	1%
Evidential difficulties, suspect identified: victim does not support further action (ratio)	6%	11%	13%	16%	18%	20%
Evidential difficulties, suspect not identified: victim does not support further action (ratio)	1%	3%	4%	5%	5%	5%

Source: Home Office (2020b)

Figure 4 and Table 3 show the distribution of 43 forces’ year-on-year changes in ‘victim does not support – suspect identified’ outcomes between 2016 and 2020.^{[footnote 27]} Several features are worth highlighting. First, the median year-on-year change for this outcome was at its highest when comparing years ending March 2017 to 2018 at 42%. This compared with lower median increases in the years either side (see Table 3). Second, compared to other years, year ending March 2018 saw more forces that more than doubled their ‘victim does not support – suspect identified’ outcome (five forces, with year-on-year increases ranging from 101% to 162%). This compares with three forces in years ending March 2016 to 2017 and one in each of the later comparison periods. Of the five forces which recorded a doubling in this outcome, all but one recorded arrest to PCB ratios of below 5% in the year ending March 2018.^{[footnote 28]}

This analysis does not, of course, provide direct evidence that the PCB reforms caused the increase in ‘victim does not support – suspect identified’ outcomes in some forces in the immediate aftermath of the reforms. As noted above, many factors can result in victims withdrawing their support for the investigation. However, the data presented are consistent with a theory that the reduction in PCB may have contributed in part to the uplift in ‘victim does not support’ volumes in some forces in the year after reform.

Figure 4: Percentage change in ‘victim does not support – suspect identified’, by force area and year, years ending March 2016 to 2020(The figures in this chart relate to outcomes recorded in the year regardless of when the associated crime was recorded.Excludes GMP.)

Source: Home Office (2020b)

Table 3: Percentage change in ‘victim does not support – suspect identified’, by force area and year, years ending March 2016 to 2020(The data in this table relate to outcomes recorded in the year regardless of when the associated crime was recorded.Excludes GMP.High outliers in a box and whiskers plot are larger than Q3 by at least 1.5 times the interquartile range.)

Years ending 31 March	2016/17	2017/18	2018/19	2019/20
Median	30%	42%	30%	12%
Range	2% to 196%	-25% to 162%	-13% to 244%	-19% to 272%
‘High’ outliers(c) (number & range)	2 (162% to 196%)	5 (101% to 162%)	1	3 (79% to 272%)
Number of forces showing increases over 100%	3	5	1	1

5.6 Victims: secondary benefits

It has been suggested that there may be some wider benefits from bail conditions, particularly in domestic abuse cases. In their update inspection on domestic abuse, HMICFRS (2019) considered the impact of the bail reforms on domestic abuse cases. Drawing on a focus group with nine practitioners from Women’s Aid, the HMICFRS inspection identified the following areas of concern:

Some police officers had been advising victims to apply for a non-molestation order incases where they had not used bail, thereby placing the responsibility on the victim toprotect themselves.

The absence of bail conditions made it hard to justify keeping the suspect away fromthe home they shared with the victim. Some suspects had been released with no bailconditions and had returned to the family home as they shared a joint tenancy with thevictim, forcing victims and their children to leave.

Where housing departments were asking for proof that a victim needed emergencyhousing, bail conditions would have previously assisted in providing this evidence.This was also found in research by Learmonth (2018) into the use of bail in rapecases where interviewees highlighted conditions are used as a mechanism toleverage safeguarding across partner agencies.

Victims were finding it harder to evidence the need for other safeguarding mechanisms – such as a restraining order – without information on bail history or breaches of bail. Additionally, Hunter, Burton and Trinder (2020) noted that in child arrangement proceedings where domestic abuse is raised, evidence of action in the CJS might be required to support a case in the family courts. Of these four examples, all but the first might be assessed as being closer aligned to secondary – or indirect – benefits of bail with conditions. In other words, it is not the bail which directly changes offender behaviour or victim’s perceptions of safety, rather it is how the existence of bail conditions may be utilised to support decisions around other remedies or protections. And while it can be questioned as to whether bail conditions are the correct mechanism for flagging risk to other agencies, the focus here is on trying to understand the possible impacts of the reforms on existing processes.

5.7 The impact on investigative management

The second area of concern around the impact of the PCB reforms is on investigative management. The main criticism here was that the creation of the presumption against bail removed a pivotal mechanism in effective investigative management. Research undertaken prior to the reforms suggested that PCB acted as a useful management tool and that the process of re-bailing suspects was helpful to effectively manage an investigation (Hillier and Kodz, 2012; Hucklesby, 2015). For instance, Hillier and Kodz (2012) found that re-bailing was particularly useful in cases where the investigation had not progressed as quickly as expected, further evidence was awaited or if the initial bail period was just not long enough.

It is argued that the shift to RUI – and the resulting absence of milestones that existed under PCB – has weakened both the discipline of case management and the supervision of suspects (HMICFRS/HMCPSI, 2020; Wiles, 2020). This in turn is thought to have contributed to longer elapsed times between arrests and case outcomes. In the most extreme cases, it has been suggested that delays have affected a small number of offences where investigations are constrained by time limits for prosecution, resulting in cases being dropped (HMICFRS/HMCPSI, 2020). The wider consequences of delays are the loss of more supportive victims and witnesses, and potentially fewer charges (HMICFRS/HMCPSI, 2020; HMICFRS/BritainThinks, 2020). HMICFRS/HMCPSI (2020) also highlight that, due to delays in investigations, there are often further delays to the prosecution of cases as suspects are notified via postal requisition.^{[footnote 29]}

Concerns about the relationship between PCB reform and the timeliness of investigations have been identified as issues relevant to both victims and suspects. However, trying to isolate the PCB reforms’ effect on timeliness of investigations is difficult. The time it takes from a crime being recorded to a charge or other outcome being given has been increasing for most offence types over recent years, and the increase pre-dates the 2017 bail reforms.^{[footnote 30]} The average (median) time taken to charge suspects after initial recording has more than doubled from 14 days in the year ending March 2016 to 33 days in 2020 (Home Office, 2020c) (Figure 5). At the time of the 2017 reforms, the year-on-year change in median values for two of the most relevant (i.e. suspect identified) outcome groups – charged and victim does not support the investigation (suspect identified) – show little change between years ending March 2017 and 2018 (up one day in each case). However, given that these are England and Wales median values for all crime types, the figures will conceal more marked changes if analysed at a more granular level (see Martin, forthcoming).

Figure 5: Median time (in days) between offence recorded and outcome assigned, years ending March2016 to 2020

Year ending March	2016	2017	2018	2019	2020
Evidential difficulties (suspect identified; victim supports action)	36	39	40	45	45
Charge/summons	14	17	18	23	33
Evidential difficulties (victim does not support action	16	14	12	16	15
Investigation complete - no suspect identified	4	2	1	1	3

Source: Home Office (2020c)

Various factors are thought to have increased investigation durations this period: additional demands on the police, including more reporting of complex crimes, and the increased need to retrieve and examine digital evidence. Richardson et al. (forthcoming) showed that between 2010 and 2018, there had been a marked increase in the proportion of investigations with some element of digital evidence. And while this effect is most obvious in sexual offence cases, it is present across a wide range of crime types.

The College of Policing (2016) identified that forensic analysis was one of the key drivers behind long periods of PCB before the reforms. They found that 60% of cases over 90 days had some form of forensic samples as the reason for bail. They also found that the most frequent type of forensic analysis was ‘phone downloads’, with 33% of cases where suspects were bailed over 90 days giving this as the reason for extending bail. While ‘computer interrogation’ only accounted for 3% of cases it had the longest mean length of bail at 84 days. Some 63% of the cases where ‘computer interrogation’ was given as a reason related to rape or other sexual offences, while 10% related to fraud offences.

With digital forensic demands increasing and investigation durations extending, it has been suggested that officers simply know in advance that they will need to seek a magistrate’s extension after the initial three months’ extension to bail has elapsed. It is therefore easier for the police to release the suspect under investigation (HMICFRS/HMCPSI, 2020; Martin, forthcoming).

The College of Policing (2016) also found that longer bail periods also reflected the need to obtain a witness statement from a third party, e.g. from a medical professional. Hales and Wiggett (2017) and Martin (forthcoming) made similar observations on the factors contributing to investigative delays.

5.8 The impact on suspects

Concerns regarding the impact that the 2017 bail reforms have had on suspects have also been raised (The Law Society, 2019). There are two main dimensions to the impact on suspects. First, that the large-scale shift to RUI has led to weaker communications between investigators and suspects; and, second, specific concerns around the monitoring of RUI suspects by forces, with potential issues around the handling of biometric samples. These impacts are closely linked to the impact on investigative management. Each is dealt with in turn.

While the reforms sought to reduce the duration of individuals on PCB, data suggest that these long durations in ‘limbo’ on PCB have, because of the reforms, simply shifted to suspects being RUI. Lengthy investigations can place a strain on suspects, victims, witnesses and families (HMICFRS/HMCPSI, 2020). Moreover, the process of updating the suspect on the progress of the investigation, which was a legal obligation under PCB, is not a mandatory feature of RUI. Confusion has existed in some forces about who is responsible for providing updates to suspects about progress of the investigation (HMICFRS/HMCPSI, 2020). Under RUI, it has been suggested that suspects are simply not informed of the progress of the investigation (The Law Society, 2019).

The Biometrics Commissioner also identified some specific weaknesses around the PCB reforms and its impact on case management, and specifically around suspect management and the handling of forensic samples. In his 2019 annual report, it was noted that some forces could not provide data about the number of suspects RUI due to weaknesses in force monitoring of these cases. This was believed to indicate the problem faced by most forces that IT systems could not be adapted to record and monitor suspects released, unless they were on bail. HMICFRS/HMCPSI (2020) have also expressed concern about the lack of a recording system for RUI in many forces. This further raised issues around the management of biometric samples. A person with no other convictions where NFA was taken against them would normally, by law, have their biometrics deleted at this point if they were not held on bail. But because many forces’ IT systems had not been rapidly modified, the biometrics of suspects who are RUI could be being held unlawfully and as a result could produce unlawful forensic matches. Although progress was being made by forces (see HMICFRS/HMCPSI, 2020), the Commissioner was critical of the delays in the system updating two years after the PCB reforms (Wiles, 2020).

5.9 Voluntary attendance and pre-charge bail

Finally, it is important to touch briefly on the relationship between the PCB reforms and so-called voluntary attendance as an alternative to arrest. Prior to the 2012 changes to Code G of PACE, suspects being investigated by the police for criminal offences were, when sufficient grounds were present, arrested for criminal offences. Since the revised Code G was introduced, the use of arrest has declined. Suspects who are not arrested may be asked to attend voluntarily, at a specific time and place (usually at a police station), to answer police questions. This is called voluntary attendance (VA) or voluntary interviews. It has been estimated that, overall, around one-third of those investigated as suspects are now subject to VA, rather than being arrested (Wiles, 2020).

The precise nature of the interaction between the growth in VA numbers and the PCB reforms is difficult to disentangle. Growth in the use of VA has been criticised by victims’ groups (e.g. The Centre for Women’s Justice, 2019) as, prior to the reforms, its use was likely to have independently led to reductions in using bail (since bail could only be granted to arrestees). And others have been critical of the increased use of VA for offences which might be expected to yield an arrest – sexual offences and violent offences – because they preclude the taking of biometric samples pre-charge (Wiles, 2020). National figures on VA by offence type are limited but ONS has published figures from HMICFRS on arrests and VAs for domestic abuse-related crimes in the year ending March 2019 (ONS, 2019). For 27 forces that provided arrest and VA data, there were approximately 150,300 arrests and 21,300 VA for domestic abuse-related crimes in the year ending March 2019 (ibid.). It has been suggested that the reduction in the use of bail with conditions post-reform has made VA appear relatively more attractive compared with arrest as the advantages of arrest over VA diminish if the ability to apply bail with conditions is less commonly available (Wiles, 2020). Increased use of VA may also have other consequences on suspects, victims and investigative management. However, data quality on VA is not sufficient to assess this.

6. Concluding remarks

This paper has examined a range of evidence around the use of PCB and the impact of the PCA 2017 reforms. It has been undertaken to support the Government review of PCB. The evidence and data reviewed point to PCB having three potential main impacts. First on investigative management, allowing police investigative processes to operate efficiently to obtain evidence, with clear checkpoints at regular intervals on case progression. Secondly, and closely linked to the first, it provides a route to keep up information flows to maintain contact with suspects under the duty to re-attend at a police station. Thirdly, it provides a vehicle to impose conditions on suspects for the protection of witnesses, victims as well as suspects, and to discourage suspects from re-offending, both generally, and specifically in relation to the original complainant.

The evidence base on PCB is limited. Unlike some police processes in England and Wales which are covered by robust, routine, nationally-collected statistics, these are not available for PCB. PCB functions as a policing power within the investigative process – it is not a targeted intervention. This might partly explain why it has not been subject to extensive evaluative research. In addition, the variation in international domestic legal systems’ treatment of suspects means that there is no pool of comparative international research literature to draw upon. The PO evidence is related, but arguably not closely enough to allow those findings to be easily transferred. Finally, the reforms to PCB have taken place against a backdrop of considerable change in the nature of recorded crime, and how the police respond to it, making any assessment of change from the available data challenging.

In the absence of more robust material, this evidence review has drawn on the available patchwork of research and data. Some elements are clearer than others. In the period after the 2017 reforms, PCB volumes fell rapidly and have recovered more recently to some degree. Variations at force level in PCB practice appear to have been considerable before the reforms, and inasmuch as the data permit, variations continue to exist. The limited evidence on victims’ perceptions of the benefits of PCB as a source of support in close relationship offences appear positive. Police and practitioner concerns exist over the loss of clear checkpoints in RUI and perceptions of a general weakening of investigative management. And there are suggestions that, along with other pressures such as the long-term growth in slow-to-process digital evidence, changes to PCB have contributed to longer investigations. The move towards RUI has brought with it a new set of concerns to suspects around reduced levels of contact over progress of investigations.

7. Bibliography

Bail Act (1976) Available at: https://www.legislation.gov.uk/ukpga/1976/63/contents [viewed 7 Jan 2021]

Benitez, C., McNeil, E. and Binder, R. (2010) ‘Do Protection Orders Protect?’ Journal of the American Academy of Psychiatry and the Law Online, vol. 38(3) pp. 376-85

Cape, E. (2016) What if police bail was abolished? The Howard League for Penal Reform [viewed 7 Jan 2021]

College of Policing (2014) Response to the Consultation on the Use of Pre-Charge Bail: Improving Standards for the Police Forces of England and Wales [viewed 7 Jan 2021]

College of Policing (2016) Bail report: Pre-charge bail – an exploratory study [viewed 7 Jan 2021]

College of Policing (2017) Pre-charge bail data [viewed 7 Jan 2021]

Cordier, R., Chung, D., Wilkes-Gillan, S. and Speyer, R. (2019) ‘The Effectiveness of Protection Orders in Reducing Recidivism in Domestic Violence: A Systematic Review and Meta-Analysis’. Trauma, Violence & Abuse. Sage Publishing, USA

CPS.gov.uk (2019) Bail [viewed 7 Jan 2021]

Criminal Justice Act (2003) Available at: https://www.legislation.gov.uk/ukpga/2003/44/contents [viewed 7 Jan 2021]

Dowling, C., Morgan, A., Hulme S., Manning, M. and Wong, G. (2018a) ‘Protection orders for domestic violence: A systematic review’. Trends & issues in crime and criminal justice. No. 551. Canberra: Australian Institute of Criminology

Dowling, C., Morgan, A., Boyd, C. and Voce, I. (2018b) Policing domestic violence: A review of evidence. Research Report no. 13. Canberra: Australian Institute of Criminology

Feist, A., Ashe, J., Lawrence, J., McPhee, D. and Wilson, R. (2007) Investigating and detecting recorded offences of rape. Home Office Online Report 18/07 [viewed 7 Jan 2021]

Gov.uk (2019) Government announces plans to review pre-charge bail law [viewed 7 Jan 2021]

Hales, G. and Wiggett, I. (2017) Pre-charge bail reforms: Context and consequences [viewed 7 Jan 2021]

Hillier, J. and Kodz, J. (2012) The police use of pre-charge bail: An exploratory study National Policing Improvement Agency [viewed 7 Jan 2021]

HMIC (2015) Increasingly everyone’s business: A progress report on the police response to domestic abuse [viewed 7 Jan 2021]

HMICFRS and BritainThinks (2020) Research into victim and suspect experiences of changes to the Policing and Crime Act 2017 [viewed 7 Jan 2021]

HMICFRS and HMCPSI (2020) Pre-charge bail and released under investigation: Striking a balance [viewed 7 Jan 2021]

HMICFRS (2018) PEEL: Police effectiveness 2017 [viewed 7 Jan 2021]

HMICFRS (2019) The police response to domestic abuse: An update report [viewed 7 Jan 2021]

Holt, V., Kernic, M., Wolf, M., & Rivara, F. (2003) ‘Do protection orders affect the likelihood of future partner violence and injury’. American Journal of Preventive Medicine, vol 24(1), 16-21

Home Office (2014) Pre-charge bail: A Consultation on the Introduction of Statutory Time Limits and Related Changes [viewed 7 Jan 2021]

Home Office (2015) Pre-Charge Bail. Summary of Consultation Responses and Proposals for Legislation [viewed 7 Jan 2021]

Home Office (2016) Domestic Violence Protection Orders (DVPO): One year on – Home Office assessment national roll-out [viewed 7 Jan 2021]

Home Office (2018) Police powers and procedures, England and Wales year ending 31March 2018 [viewed 7 Jan 2021]

Home Office (2019) Police powers and procedures, England and Wales year ending 31 March 2019 [viewed 7 Jan 2021]

Home Office (2020a) Police powers and procedures, England and Wales year ending 31 March 2020 [viewed 7 Jan 2021]

Home Office (2020b) Police recorded crime and outcomes open data tables [viewed 7 Jan 2021]

Home Office (2020c) Crime outcomes in England and Wales 2019 to 2020 [viewed 7 Jan 2021]

Hotaling, G. and Buzawa, E. (2003) Victim Satisfaction with Criminal Justice Case Processing in a Model Court Setting. Department of Criminal Justice. University of Massachusetts [viewed 7 Jan 2021]

House of Commons Home Affairs Committee (2015) Home Affairs – Seventeenth Report: Police bail [viewed 7 Jan 2021]

Hucklesby, A. (2001) ‘Police Bail and the Use of Conditions’. Criminal Justice, vol. 1(4), pp.441-463

Hucklesby, A. (2015) Pre-charge bail: an investigation of its use in two police forces [viewed 7 Jan 2021]

Hucklesby, A. and Marshall, E. (2000) ‘Tackling Offending on Bail’. The Howard Journal of Criminal Justice, vol. 39(2), pp.150-170

Hunter, R., Burton, M. and Trinder, E. (2020) Assessing Risk of Harm to Children and Parents in Private Law Children Cases (Final Report). Ministry of Justice [viewed 7 Jan 2021]

Kelly, L., Adler, J., Horvath, M., Lovett, J., Coulson, M., Kernohan, D., Gray, M., Hillier, J. and Nicholas, S. (2013) Evaluation of the Pilot of Domestic Violence Protection Orders[7 Jan 2021]

Kothari, C., Rhodes, K., Wiley, J., Fink, J., Overholt, S., Dichter, M., Marcus, S. and Cerulli, C. (2012) ‘Protection Orders Protect Against Assault and Injury’. Journal of Interpersonal Violence, vol. 27(14), pp.2845-2868

Learmonth, S. (2018) A system to keep me safe: An exploratory study of bail use in rape cases. Dissertation: London Metropolitan University [7 Jan 2021]

mygov.scot (2020) Being arrested: your rights [viewed 7 Jan 2021]

Logan, T. and Walker, R. (2009) ‘Civil Protective Order Outcomes’. Journal of Interpersonal Violence, vol. 24(4), pp.675-692

Logan, T., Cole, J., Shannon, L., & Walker, R. (2007) ‘Relationship characteristics and protective orders among a diverse sample of women’. Journal of Family Violence, vol. 22,pp.237–246

Logan, T., Shannon, L., & Cole, J. (2007) ‘Stalking victimization in the context of intimate partner violence’. Violence and Victims, vol. 22(6), pp. 669–683

Logan, T., Walker, R., Shannon, L. and Cole, J. (2008) ‘Factors Associated with Separation and Ongoing Violence among Women with Civil Protective Orders’. Journal of Family Violence, vol. 23(5), pp. 377-385

Martin, R. (forthcoming) Pre-Charge Bail Research Project: Summary of Descriptive Findings

McFarlane, J., Malecha, A., Gist, J., Watson, K., Batten, E., Hall, I., et al. (2004) ‘Protection orders and intimate partner violence: An 18-month study of 150 black, Hispanic, and white women’. American Journal of Public Health, vol. 94(4), pp. 613–618

Morgan, P. and Henderson, P. (1998) Remand Decisions and Offending on Bail. London: Home Office

NPCC (2019) Police chiefs guide officers to impose bail conditions protecting victims and vulnerable people [viewed 7 Jan 2021]

Office for National Statistics (ONS) (2019) Domestic abuse and the criminal justice system – Appendix tables [viewed 7 Jan 2021]

Office for National Statistics (ONS) (2020) All data related to Domestic abuse and the criminal justice system, England and Wales: November 2020 [viewed 7 Jan 2021]

Phillips, C. and Brown, D. (1998) Entry into the criminal justice system: a survey of police arrests and their outcomes. Home Office Research Study (185). Home Office, Research and Statistics Directorate, London, UK

Police and Criminal Evidence Act (PACE) (1984) Available at: https://www.legislation.gov.uk/ukpga/1984/60/contents [viewed 7 Jan 2021]

Policing and Crime Act (2017) Available at: https://www.legislation.gov.uk/ukpga/2017/3/contents/enacted [viewed 7 Jan 2021]

Sleath, E., and Smith, L. L. (2017) ‘Understanding the factors that predict victim retraction in police reported allegations of intimate partner violence’. Psychology of Violence, vol. 7(1), pp. 140-149.

Raine, J. and Willson, M. (1995) ‘Just Bail at the Police Station?’ Journal of Law and Society.Dec.1995, Vol. 22(4), pp.571-585

Raine, J. and Willson, M. (1997) ‘Police bail with conditions: Perspectives on the Use, Misuse and Consequences of a New Police Power’. British Journal of Criminology. Vol. 37(4), pp. 593-607

Richardson, V.et al. (forthcoming) Assessing the increase in digital evidence in police investigations from appeal case summaries

The Centre for Women’s Justice (2019) Super-complaint: police failure to use protective measures in cases involving violence against women and girls [viewed 7 Jan 2021]

The Law Society (2019) Release under investigation [viewed 7 Jan 2021]

Wiles, P. (2020) Commissioner for the retention and use of biometric material. Annual Report 2019 [viewed 7 Jan 2021]

Appendix A. Information from comparative jurisdictions

The Northern Ireland (NI) Law Commission conducted a consultation into bail in 2010 and published the findings in a report in 2012 (NI Law Commission, 2010; 2012). It found that, at the time of writing, only two jurisdictions had the police power to grant bail before charge – England and Wales and NI. This is because it was viewed as against suspects’ rights to restrict their civil liberties when they had not been charged with an offence (ibid.). The review recommended that PCB be reformed due to the ‘complexity, uncertainty and inconsistency’ of the system, particularly the discrepancies between police and court bail (ibid.).

The NI Law Commission recommended:

• the removal of the power to attach conditions to PCB
• the removal of the duty to surrender to custody and the offence of failure to surrender, the imposition only of a requirement to attend a police station
• the creation of a right to have the decision to release on bail reviewed.

However, the NI Law Commission became non-operational before any of its recommendations could be taken forward. Thus, recommendations around PCB were never actually implemented and the Police Service of Northern Ireland (PSNI) continues to use PCB as before. Bail conditions are more limited in NI than England and Wales (PSNI, 2019).

Scotland passed the Criminal (Scotland) Justice Act (2016), which introduced the power of ‘investigative liberation’. This power allows the police to liberate a suspect from custody without charge where further enquires are required. Conditions may be applied to this release, similar to PCB, such as prohibiting individuals from certain places (mygov.scot, 2020).

Appendix B. Available data on volumes of PCB pre-and post-reforms

Source	Coverage Time period Number put on PCB % of those arrested released on PCB	Time period	Number put on PCB	% of those arrested released on PCB
Home Office (2020)	40 police forces in England and Wales	1 Apr 2019 to 31 Mar 2020	153527	N/A
Home Office (2019)	41 police forces in England and Wales	1 Apr 2018 to 31 Mar 2019	86628	N/A
The Law Society (2019)	29 police forces in England and Wales	3 Apr 2017 to 2 Apr 2018	32283	N/A
The Law Society (2019)	29 police forces in England and Wales	3 Apr 2016 to 2 Apr 2017	203228	N/A
Home Office (2018)	17 police forces in England and Wales	1 Apr 2017 to 31 Mar 2018	16491	N/A
College of Policing (2017)	30 police forces in England & Wales	Mar 2017 (one month)	13452	0.26
College of Policing (2016)	26 police forces	Apr 2013 to Mar 2014	300712	0.31
Home Office (2015) – original data from College of Policing	12 police forces provided data and then figures were scaled up(a)	Apr 2013 to Mar 2014	404000	N/A
House of Commons Home Affairs Committee (2015)	Not given	N/A	303000	0.31
Home Office (2014) – original data from the BBC	40 police forces but scaled up by the HO	2013 to 2014	78679	N/A
ACPO/NPIA data collection exercise (Hillier & Kodz, 2012)	24 forces in England & Wales provided some data, with 12 forces returning all the information requested	Oct 2010 to Mar 2011 (six months)	N/A	33% of those taken into custody released on PCB

These figures have been scaled nationally through the number of arrests for notifiable offences per force, according to the Police Powers and Procedures E&W 2012/13 publication.

Appendix C. Additional graphs on the use of PCB and RUI

Figure 6: Number of individuals put on PCB before and after the reforms (years ending 2 April2017 and 2018) for selected forces(This chart excludes data from London Metropolitan Police. Approximate figures)

Force	2016/17	2017/18
Avon and Somerset	7255	367
Bedfordshire	3317	514
Cambridgeshire	5330	824
Cheshire	4742	526
Cleveland	1693	78
Cumbria	3877	614
Derbyshire	658	159
Devon and Cornwall	4743	211
Dorset	3222	87
Essex	14441	3954
Greater Manchester	12370	1846
Gwent	2950	366
Hertfordshire	1224	77
Lancashire	843	1115
Leicestershire	4786	253
Lincolnshire	3665	437
Merseyside	3874	929
Norfolk	530	889
North Wales	4247	447
Northamptonshire	2949	331
Nottinghamshire	7392	562
South Wales	5333	2869
Staffordshire	5628	463
Suffolk	867	832
Surrey	5280	1033
Thames Valley	13768	379
Warwickshire	2865	543
West Mercia	7541	1697

Source: The Law Society, 2019

Figure 7: Estimated ratio of PCB volumes to arrests before and after the reforms in 2016/17 and 2017/18 for 28 forces

(There is a two-day difference between the timeframes for the arrest data collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by the Law Society (3 April to 2 April).)

Source: The Law Society, 2019; Home Office, 2020a

Appendix D. Exploring the wider evidence base: post-charge bail, court bail and POs

D.1 Research on effectiveness of post-charge or court bail

Research on effectiveness of post-charge bail or court bail largely focuses on whether conditions were breached and offending on bail. Raine and Wilson (1995) found in their interviews with defendants that there was a variety of reactions to particular conditions attached to their post-charge bail. Of 79 people interviewed, a majority (54%) stated that they complied with their conditions. 44% indicated that their conditions had ‘helped them keep out of trouble’. Interviews also revealed that defendants realised that more generalised conditions such as exclusion from town were not well enforced. Over half the sample indicated that they knew (or learned by the end of their bail period) not to expect enforcement by the police if they were caught breaking their conditions.

Hucklesby and Marshall (2000) investigated the impact of Section 26 of the Criminal Justice and Public Order (CJ&PO) Act 1994 on magistrates’ remand decisions which removed the presumption of bail for defendants who have allegedly committed offences on bail. They analysed data for periods before and after the implementation of the Act in one of the courts used by the Bail Process Project^{[footnote 31]} (Leicester magistrates’ court). They found that the provisions of bail had little practical effect on the number of defendants who had allegedly committed offences on bail who were then remanded to custody. They did, however, identify an increase in the number of such defendants who were granted bail with conditions. Changes were found in remand decisions for two groups of defendants: those charged with serious offences who already had a bail history, and defendants charged with vehicle crime and burglary.

Morgan and Henderson (1998) evaluated the Bail Process Project. The project ran in five court areas and data were collected in 1993 and 1994. Different measures and initiatives were introduced in the different areas. The research indicated that the project did show some positive effects with two areas (Horseferry Road area and Leicester), which showed a reduction in rates of offending on bail. However, the other three showed little change. They attributed the reduction in the Horseferry Road areas to more persons being remanded in custody (custody rates increased by six percent over the same time period). The reduction in Leicester was attributed to improvements made such as increased emphasis on training of magistrates. The study also identified some characteristics of defendants who had higher rates of offending while on bail. These included: no fixed abode, charged with car theft or burglary, aged 17 and under, unemployed, those who waited more than 3 months before their trial, and those with a previous custodial sentence or record for breaching bail. Lower rates of offending on bail were found for those who were included: in employment, waited less than a month before trial, were charged with assault or fraud, and were aged 21 or over.

D.2 The evidence base on the effectiveness of protection orders

Protection orders (POs) are civil law remedies that aim to provide protection for victims and protect the public from further potential victimisation. POs can either be applied for by the victims themselves to the courts, the police on behalf of the victims or the court on behalf of the victim. Compared with PCB, which can be imposed for any offence, POs are typically only given for certain offence types (or example, in England and Wales there are DVPOs which are given for domestic violence offences). There are a wide variety of POs ranging from those that aim to protect against domestic violence to gang injunctions (see Appendix E).

PCB and POs share some similar characteristics. For example, both:

• Place certain restrictions on individuals e.g. no contact with victim or an exemption from entering certain locations or addresses.
• Both aim to provide protection to witness(es) and victim(s) and are widely used to reduce harm to those who are at risk.
• In the same way that PCB conditions are applied by the police, some POs can be applied for by the police – e.g. DVPOs.

While PCB and POs bear some similarities, in other respects, the differences between them are considerable. In particular, (a) in some jurisdictions, violations (equivalent to breaches of bail) can result in a punitive response (b) duration of the PO is typically longer than PCB (c) POs tend to have limited coverage in terms of specific offence types and (d) POs are usually victim-generated rather than an automatic CJS response initiated by the police. These differences ultimately make it difficult to directly translate evidence on the effectiveness of POs across to PCB. It remains unclear what active ingredient makes POs effective in some circumstances and for some victims. In particular, it is difficult to unpick the potentially pivotal nature that tougher sanctions may play in the effectiveness of POs.

The evidence base on the effectiveness of POs suggests that there are some situations where POs may be effective in reducing re-offending and protecting the victim. To assess the comparability between POs and PCB, four ‘core’ studies, which featured across two systematic reviews, were subject to more detailed examination (Dowling et al., 2018a; Cordier et al., 2019). These studies were undertaken by Kothari et al. (2012), McFarlane et al. (2004), Holt et al. (2003) and multiple papers by TK Logan and colleagues that drew on the same sample (Logan, Cole, Shannon and Walker, 2007; Logan, Shannon and Cole, 2007; Logan et al., 2008; Logan and Walker, 2009). These ‘core’ studies are outlined in Appendix F. A study by Kelly et al. (2013) of DVPOs in England and Wales is also reported on as it is one of the few UK-based studies.

D.3 Re-offending

Systematic reviews (Dowling et al., 2018a; Cordier et al., 2019) provide some evidence that POs can be effective at reducing recidivism. Dowling et al. (2018a) undertook a meta-analysis of four studies and found that POs were associated with a small but statistically significant overall reduction in severe domestic violence re-victimisation. Three of these studies were based in the US (Holt et al., 2003; McFarlane et al., 2004; Kothari et al., 2012). All of these studies used control groups to test the effect of POs in intimate partner violence (IPV) cases. Holt et al. (2003) found that POs were associated with a decreased likelihood of subsequent physical and non-physical IPV. Similarly, based on interviews with 150 women who had applied for a PO in Texas, McFarlane et al. (2004) found that women reported significantly lower levels of IPV and harassment up to 18 months after applying for a PO. Kothari et al. (2012) measured the efficacy of POs in reducing assault and injury-related outcomes in the US. They found that IPV victims with POs had significantly fewer emergency department visits and significantly fewer police incidents after a PO than before.

In the context of England and Wales, Kelly et al. (2013) found that DVPOs were associated with reduced rates of victimisation. In their evaluation of the DVPO pilot in three forces, they found that DVPOs were most effective in reducing re-victimisation for ‘chronic cases’ (ibid.). The authors used police data and call-outs to measure revictimisation, although these may not capture the full extent of re-victimisation.

Some robust studies found that some victims continued to suffer violence or violations. Logan and Walker (2009) found that 58% of women experienced PO violations and Kothari et al. (2012) found that nearly half (49%) of women with PO orders reported violations.

Kelly et al. (2013) found that less than 1% of DVPOs were reportedly breached. After DVPOs were rolled out nationally in England and Wales in 2014, HMIC (2015) reported that 17% of DVPOs that had been granted by the courts had been breached. Similarly, in their year assessment of DVPOs, the Home Office (2016) reported that 18% of DVPOs had been breached since implementation. These findings indicate that there is likely to have been underreporting of breaches in Kelly et al. (2013)’s study. Moreover, on the basis of other research, it is likely that data on breaches recorded by the police underreports the actual level of breaches. Hotaling and Buzawa’s study (2003) found around half of violations were reported to the police.

Some groups appeared to be at higher risk of committing a PO violation – offenders with a history of stalking, violence or criminal acts, and victims with lower socioeconomic status (Logan and Walker, 2009; Benitez et al., 2010; Cordier et al., 2019).

D3.1 Victim perceptions of effectiveness and safety

Focusing on violations may not capture all the potential benefits of POs. Cordier et al. (2019) highlighted that simply using recidivism to measure the effectiveness of POs cannot account for all the complexities involved in reducing domestic abuse. A victim’s sense of safety and perceived effectiveness are also important to consider. Applying for a PO made some victims feel as though something was being ‘done’, and the issuing of POs may have a more positive impact on victim-police interaction. These perceived benefits have also been identified from qualitative research with victims whose cases involved suspects receiving conditions under PCB (see HMICFRS and BritainThinks 2020).

Logan and Walker (2009) examined two dimensions of PO effectiveness – violations of POs and women’s perceptions of PO effectiveness and safety – across four jurisdictions in the US. They also explored the factors most associated with these perceptions. Overall, they found the women perceived POs to be extremely (51%) or fairly (27%) effective, and they reported feeling extremely (43%) or fairly (34%) safe. Around one in five (22%) did not feel they were effective and did not feel safe even with a PO. Using the same sample of women, Logan et al. (2007) found that although most of the women in their study reported the PO as ‘effective’, almost one in four women reported that their partner had violated the order. This suggest that victims can perceive the PO as effective even if violence continues, and highlights the challenges of solely assessing effectiveness based on rates of re-victimisation.

In their evaluation of pilot DVPOs in England and Wales, Kelly et al. (2013) conducted interviews with victim-survivors. They found that DVPOs were generally perceived positively, though a minority of victims did not feel DVPOs were useful. However, these findings are based on a small sample size of victim-survivors (16 victim-survivors were originally interviewed, with nine completing follow-up interviews). Additionally, the majority of victims interviewed had received support from support services alongside DVPOs. This can skew the views of how effective DVPOs are, as only two-thirds of the pilot received support services alongside. These findings imply that DVPOs are likely to be more effective when offered alongside support services.

Victim perspectives have played a key role in the criticisms surrounding the 2017 reforms (CWJ, 2019; Learmonth, 2018). Research undertaken by Logan and Walker (2009) indicated that the majority of women in receipt of a PO perceived them as safe and effective. So even when violence continued, following the issue of the PO, victims valued the increased victim-police interaction that came as a result of a PO. This is important to PCB policy as even the issuing of bail conditions may make the victim feel safe. This links in with research on PCB use in rape cases (Learmonth, 2018). Victims felt validated when the police issued bail conditions as it made their case credible, and they felt that the police believed them.

Appendix E. Civil protection orders (CPOs) and injunctions in England & Wales

CPO/ Injunction	Who applies?	Which court?	Duration of order	Given on conviction?	Outcome if breached	Summary
Domestic Violence Protection Order (DVPO)	Police officer applies to court on behalf of victim	Magistrates	28 Days	No	Breach of DVPO is a civil breach of court order. Penalty for a breach of a civil order is £50 for every day that the person is in default of the order, up to a maximum of £5000 or 2-months imprisonment.	Domestic Violence Protection Notice (DVPN) given by the police to a personwho they reasonably believe has been violent against victim lasts 48 hours. In this time, police apply to Magistrates’ Court for DVPO which can place restrictions on a person for 28 days e.g. no contact or entering address.
Non-Molestation Order (NMO)	Victim of domestic violence (DV) themselves applies to court	Family	Normally issued for a specified period. There is no limit on how long NMOs can be extended.	No	The complainant can either call the police to have the breach dealt with within the criminal jurisdiction, or they can apply to have the person committed to custody for contempt application in the civil jurisdiction. The two jurisdictions are exclusive, and prosecutors will not be involved in civil proceedings. Penalties for a breach range from fines, community orders and custodial sentences. Maximum 5 years’ imprisonment.	An NMO is aimed at preventing your partner or ex-partner from using or threatening violence against you or your child, or intimidating, harassing or pestering you, in order to ensure the health, safety and wellbeing of yourself and your children.
Occupation Order (OO)	Victim of DV themselves applies to court	Family	Normally issued for a specified period. There is no limit on how long OOs can be extended.	No	Breach of order is not a criminal offence unless power of arrest attached to order. Breaching an OO with a power of arrest could result in up to 2 years in prison or a large fine.	An OO regulates who can live in the family home, and can also restrict your abuser from entering the surrounding area. Anyone who does not feel safe continuing to live with their partner, or those who left home because of violence, but want to return and exclude their abuser, can apply for an OO.
Restraining Order (RO)	Court imposes themselves	Magistrates and Crown	Can last for a specified period or for an indefinite period.	Yes	Maximum 5 years’ imprisonment. Offences range from fine, community orders and custodial sentences.	An RO may be made on conviction or acquittal for any criminal offence. ROs are intended to be preventative and protective. The guiding principle is that there must be a need for the order to protect a person or persons. An RO is therefore preventative, not punitive. The test to be applied by the court before making an order is whether an order is necessary to protect the persons named in it from harassment or conduct that will put them in fear of violence.
Stalking Protection Order (SPO)	Police or the courts	Magistrates	Minimum of 2 years up to an indefinite period (until further order).	No	Summary conviction – imprisonment not exceeding 12 months or a fine or both.Indictment – imprisonment for a term not exceeding 5 years or a fine or both.	The police should consider applying for an order where it appears to them that: 1. the respondent has carried out acts associated with stalking 2. the respondent poses a risk of stalking to a person 3. there is reasonable cause to believe the proposed order is necessary to protect the other person from that risk. (The person to be protected does not have to have been the victim of the acts mentioned above.) Magistrates’ Court can also make an order when similar criteria are met e.g. victim has reported stalking behaviour or belief victim is at risk of harm. No prior conviction for stalking offences is required to apply for an order.
Gang injunction	Chief officer of the police/ chief constable/ local authority e.g. councils	County or High	Up to 2 years	No	Breach of injunction of this type is not a criminal offence – it is dealt with as a civil contempt of court for adults and by way of a separate statutory scheme for under 18s.	Anyone seeking to apply for an injunction must have evidence that the respondent has engaged in, encouraged or assisted gang-related violence or gang-related drug dealing, and will need to prove this on the balance of probabilities at court.Applicants will also need to convince the court that the gang injunction is necessary to prevent the respondent from being involved in gang-related violence and gang-related drug dealing and/or to protect the respondent from such violence or drug dealing activity.
Criminal Behaviour Order (CBO)	Criminal courts	Magistrates, Crown, Youth	Adults: 2 years – no maximum time frame, may be indefinite Under 18s: 12 months to 3 years	Yes	It is a criminal offence to breach terms of a CBO – up to 5 years’ imprisonment or fine.	To prevent anti-social behaviour by a person convicted of any other criminal offence. CBOs can be made in addition to a sentence. They replace Anti-social Behaviour Orders (ASBOs) on conviction.
Knife Crime Prevention Order (KCPO)	Courts	Magistrates, Crown, Youth	At least 6 months but no more than 2 years	Yes	Maximum penalty for a breach is 6 months’ imprisonment or fine or both on summary conviction, or 2 years’ imprisonment, a fine or both, following conviction on indictment.	KCPOs were introduced through the Offensive Weapons Act 2019 and will give officers an additional tool to steer those most at risk away from violent crime.The civil orders can be imposed by courts on any person aged 12 or over who police believe is regularly carrying a knife, or upon conviction of a knife-related offence.
Serious Crime Prevention Order (SCPO)	Courts	Court, High	Maximum period of 5 years	Yes	Criminal offence punishable up to 5 years and unlimited fine.	The court must have reasonable grounds to believe that an order would protect the public by preventing, restricting or disrupting involvement by the person in serious crime in England and Wales.An application for a SCPO should generally only be made either following a conviction for a serious offence or following a decision that, applying the Code for Crown Prosecutors, the evidence available does not provide a realistic prospect of a conviction, or a prosecution would not be in the public interest, for reasons other than the availability of a SCPO.
Violent Offender Order (VOO)	Court makes orders made on complaint by chief office of police	Magistrates	Minimum 2 years up to 5 years (unless renewed or discharged)	No	Breach of the terms of a VOO, or failure to comply with the notification requirements of a VOO, constitutes a criminal offence punishable by 5 years’ imprisonment.	A VOO is a civil order intended to protect the public from ‘qualifying offenders’ who pose a current risk of ‘serious violent harm’.They are available for offenders over 18 who have received at least 12 months’ custody or a hospital order or found not guilty by reason of insanity for manslaughter and murder offences.
Sexual Harm Prevention Order (SHPO)	Complaint by the police or National Crime Agency (NCA), Courts	Magistrates, Crown, Youth, Court of Appeal	Minimum of 5 years and has no maximum length.	Yes	Breach of any prohibition of an order is a criminal offence, with a maximum penalty of 5 years’ imprisonment.	Made by the court for an individual convicted or cautioned, including youth cautions, for a relevant offence and who poses a risk of sexual harm to the public in UK or children or vulnerable adults abroad.Aim is to restrict the harmful behaviour of offenders convicted of a sexual or violent or other dangerous offence listed in Schedule 3 or Schedule 5 of the 2003 Act.SHPOs can be given at time of sentencing, so imposed on conviction or caution.
Sexual Risk Order (SRO)	Police	Magistrates, Youth	Minimum of 2 years to indefinite	No	Any breach of the order is a criminal offence punishable by a maximum of 5 years’ imprisonment.	Difference between this and a SHPO is that an individual does not need to have committed a relevant offence, or indeed, any offence. Civil order that can be sought by the police against an individual who has not been convicted, cautioned etc of an offence but who is none the less thought to pose a risk of harm.
Terrorism Prevention and Investigation Measures (TPIM) notice	Home Secretary	Court does review decision but it does not specify online which court this is	Maximum of 2 years	No	Criminal offence carrying maximum penalty of 5 years’ imprisonment.	Can place conditions on suspected terrorists before conviction.The court reviews whether the conditions for imposing TPIMs are satisfied.
Slavery Trafficking Prevention Order (STPO)	Courts, police, immigration officer, NCA	Crown, Magistrates, Youth	Minimum of 5 years to indefinite	Yes	Breach is criminal offence punishable by up to 5 years’ imprisonment.	STPOs can be made on sentencing by a court who is dealing with a defendant of a slavery or human trafficking offence, and on application by a chief officer in a police force, immigration officer or the Director General of the NCA. Conditions are:1. There is a risk that the defendant may commit a slavery or human trafficking offence.2. It is necessary to make the order to protect persons generally, or particular persons, from the physical or psychological harm which would be likely to occur if the defendant committed such an offence.
Slavery and Trafficking Risk Order (STRO)	Police, NCA or immigration office	Magistrates	No more than 5 years	No	Breach of STRO is punishable by up to 5 years’ imprisonment.	Order made if a defendant has not been convicted of a trafficking or slavery offence but who is nevertheless thought to pose a risk of harm and it is necessary to protect others.
Drug Dealing Telephone Restriction Order (DDTRO)	Courts, police, NCA	N/a	Can be indefinite	Yes	N/a	A DDTRO stops a phone number being used or reactivated.The police or NCA will have to satisfy a court that on the balance of probabilities the device has been used, is likely to have been used, or is likely to be used in connection with drug-dealing offences.
Community Protection Notice (CPN)	Council officers, police officers, PSCO, social landlords	N/a	A specified time period	No	Failure to comply with the terms of a CPN is a criminal offence. Sanctions include fine, fixed penalty notice (FPN), remedial orders, seizure.	Intended to deal with particular, ongoing problems or nuisances which are having a detrimental effect on the quality of life of those in the locality.
Female Genital Mutilation Protection Order (FGMPO)	Individual, local authority or any other person with the permission of the court can apply	Family	Can be indefinite	No	Maximum penalty for breaching an FGMPO is 5 years’ imprisonment.	Offer a legal means to protect and safeguard victims and potential victims of female genital mutilation (FGM).The order will have conditions to protect a victim or potential victim from FGM. This could include, for example, surrendering a passport to prevent the person at risk from being taken abroad for FGM, or requirements that no one arranges for FGM to be performed on the person being protected.
Forced Marriage Protection Order (FMPO)	Application made to the Family Court	Family	Can be indefinite	No	Breach of FMPO can be treated as a criminal offence or as civil contempt of court matter.	Purpose of protecting a person from being forced into a marriage or from any attempt to be forced into a marriage, or a person who has been forced into a marriage.
Public Spaces Protection Order (PSPO)	Councils	Local authority	No more than 3 years	No	Criminal offence to breach PSPO. A breach can be dealt with by FPN or prosecuted if they do not pay the FPN.	A local authority may make a PSPO if satisfied on reasonable grounds that two conditions are met.First condition:a) activities carried on in a public place within the authority’s area have had a detrimental effect on the quality of life of those in the locality, orb) it is likely that activities will be carried on in a public place within that area and that they will have such an effect.Second condition:a) is, or is likely to be, of a persistent or continuing natureb) is, or is likely to be, such as to make the activities unreasonable, andc) justifies the restrictions imposed by the notice.

Appendix F. Summary table of Protection Order (PO) / Domestic Violence Protection Order (DVPO) studies

Study title / Author(s)	Research aims	Scope of data	Methodological characteristics	Study’s main findings / Conclusions
Evaluation of the pilot of domestic violence protection orders Home Office/ Kelly et al. (2013)	How were DVPOs and DVPNs delivered? What were practitioner/victim/ perpetrator opinions of DVPOs? Were DVPOs effective in reducing DV? What was the economic value of the pilot scheme?	Country: UK 15-month range 3 forces (G. Manchester, W. Mercia, Wilts) Base data: potential DVPOs identified by police (n=509)	Mixed methods (surveys, focus groups, interviews)Quantity analysis based on police incident data – geared towards wider impact on victimisation and economy No engagement for offence severity or for longer-term effects No control group	Cases generally involved cohabiting partners, DVPOs generally perceived positively, though minority of victims did not feel DVPOs useful, Few DVPO breaches (<1%); potential DVPO effect of 1 fewer incident/case on average, Referral practice and data sharing inconsistent, Timing/admin constraints around DVPOs, Perception issues/confusion over DVPOs, Negative return on investment (0.23p/£1), Inconsistencies with judicial system processing of DVPOs
Protection Orders protect against assault and injury: A longitudinal study of police-involved women victims of intimate partner violence Kothari et al. (2012)	Measure efficacy of POs in reducing assault and injury-related outcomes	Country: USA, Base data: female victims of assault in year 2000 (n=993), Both PO and control group n=130, Retrospective review of police, emergency dept., family court, and prosecutor records for police-involved IPV victims	Between-group comparison (PO and non-PO), Records synthesised to create variables, Control group matched in pairs with PO group by similarity of situation (Chi-square analysis and logistic regression)	POs associated with more police calls for both felony/multiple count and non-assaultive incidents, PO group had incident rates higher than control group prior to PO, dropping to control group level during and after PO, Strongly confirms for protective effect of POs, and for reduced police and emergency dept visits both during and after PO
Protection orders and intimate partner violence: An 18-month study of 150 black, Hispanic, and white women McFarlane et al. (2004)	Comparison of types and frequencies of intimate partner violence (IPV) experienced by women before/after 2-year PO	Country: USA (Houston, TX), Base data: 4 interviews of women applying for POs (n=150), Timeframe: 18 months (Jan 2001 to June 2002)	Inter-group comparison (PO and non-PO), Demographic data forms, Severity of Violence against Women Scale (SVAWS), Stalking Victimisation Survey in tandem with Sheridan HARASS instrument, Data acquired through phone interviews	Whether or not 2-year PO granted, the act of seeking one is related to significantly lower reports of threatened abuse, physical abuse, stalking, work harassment and femicide risk factors (at 3, 6, 12 and 18 months after application), 44% of women reported violation of PO to study; only half of these were reported to police
Do protection orders affect the likelihood of future partner violence and injury? Holt et al. (2003)	Assess effect of CPO on risk of future self-reported IPV and injury	Country: USA (Seattle, WA),• 3 interviews of women applying for CPOs (n=448), Timeframe: 9 months after incident (Oct 1997 to Dec 1998)	Inter-group comparison (CPO and non-CPO), Demographic data on victims and abusers, Abuse history (Conflict Tactics Scale), Participant mental and physical health status (Center for Epidemiologic Studies (CES) Depression Scale; NET Screening Scale; Social Adjustment Scale; questions), Odds ratios used to estimate risks (for each interview independently)	Decreased risk of: unwanted contact; unwelcome calls; threats; psychological, sexual or physical abuse or injury; abuse-related medical care for women obtaining a CPO compared to those not
Civil protective order outcomes: Violations and perceptions of effectiveness Logan and Walker (2009)	Examined PO outcomes to evaluate: Whether violence to petitioner continues after PO obtained, Petitioner perceptions of PO effectiveness	Country: USA, Base data: 2 interviews of female victims who obtained PO against male partner (n=698), Timeframe: 21 months (Feb 2001 to Nov 2003), follow-up interview approx. 1 year after first 4 jurisdictions	Demographic data; Conflict Tactics Scale (six categories and violence severity index added at follow-up); Life History Calendar method to collect stalking experiences; perception scales of ‘effectiveness’ and ‘safety’, Logistic regression for PO violation factors, Chi-square and one-way analysis of variance (ANOVA) for bivariate differences	PO still in effect for 70% of victims at follow-up, 58% of women experienced PO violation (no significant difference across jurisdictions), POs perceived as extremely (51%) or fairly (27%) effective, and as extremely (43%) or fairly (34%) safe, Only 49% of PO violations reported to study were reported to police, Stalking history and continued relationship with partner identified as key factors predicting PO violation
Relationship characteristics and protective orders among a diverse sample of women Logan, Cole, Shannon and Walker (2007)	Comparisons based upon descriptive information of women with POs, Measured PO outcomes and victim perceptions of comparative: effectiveness; freedom from partner; feelings of safety	Country: USA, Base data: interview of female victims who obtained PO (n=757) , Split into groups of: rural White (n=371); urban White (n=254); urban African American (n=103) , Timeframe: 21 months (Feb 2001 to Nov 2003)	Demographic data; Conflict Tactics Scale; Psychological Maltreatment of Women Inventory (PMWI) , Subscales generated: 5 psychological abuse, 2 physical abuse, 1 sexual insistence, 1 sexual, 1 injury	PO violation rate 25% to 30% in all three groups, Perceptions of PO: 55% to 62% fairly or extremely safe; 56 to 65% fairly or extremely free; 70 to 80% DVO effective; 83% to 94% DVO process as good (all lowest scores in rural White group)
Stalking victimization in the context of intimate partner violence Logan, Shannon and Cole (2007)	Comparisons between women with PO against (ex-) partner who reported and not reported stalking	Country: USA, Base data: women with CPO (n=757), split into those who reported partner stalking in the past year (n=359); those who reported no previous partner stalking (n=345); and those who reported partner stalking but that it had not occurred in the last year (n=53); this last group were not included, Timeframe: 21 months (Feb 2001 to Nov 2003), 1interview (5 weeks post-PO)	Inter-group comparison; demographics taken, Stalking measured by: ‘Has partner ever repeatedly followed/ phoned you, showed up at house, work/other?’ and ‘Did your partner ever stalk or obsessively pursue you when you did not want him to, and it frightened you?’, Conflict Tactics Scale; PMWI; pilot work questions (9 victimisation subscales, 4 comparative indices), Adjusted MINI responses (mental health subscales), ‘Logan method’ for perceptions of PO scales, MANCOVA & logistic regression; findings significant at p<0.01	Partner stalking before PO strong indicator of: more severe partner violence/victimisation; increased distress; increased fear; more PO violations, all after controlling for demographics etc.,Small but significantly larger number of women without stalking before PO still living with partner at time of post-PO interview, Rare for women to report stalking spontaneously in either PO petition or for police to mention it in reports
Factors associated with separation and ongoing violence among women with civil protective orders, Logan, Walker, Shannon and Cole (2008)	Understanding relationship status after receiving PO against male partner, and factors associated with PO violations	Country: USA, Base data: women obtaining PO against male intimate partner (n=756, of whom 698 provided full data to end of study), 4 jurisdictions (3 rural, 1 urban), Timeframe: 21 months (Feb 2001 to Nov 2003), 2 interviews; first at 40 days (avg.) after PO, second at approx. 1 year after PO	Demographics taken, Relationship characteristics derived through measures derived from pilot work, Victimisation scales derived through weighting responses to Conflict Tactics Scale, ‘Logan method’ for perception scales, Chi-squares; 1-way ANOVA; logistic regression	Roughly 5 in 10 women who did not continue relationship experienced PO violation, as opposed to 7 in 10 who did continue relationship, Regardless of relationship status, majority of women felt both safer, and that PO remained effective, Stalking is significant indicator of likelihood that: relationship terminates; PO violated (regardless of post-PO relationship status), Longer relationships likelier to end after PO obtained, 40% reported PO violation rate to study

1. There is a two-day difference between the timeframes for the experimental PCB statistics collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by the Law Society (3 April to 2 April). ↩

2. There is a two-day difference between the timeframes for the arrest data collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by the Law Society (3 April to 2 April). ↩

3. The pre-conditions for PCB are that the custody officer is satisfied that releasing the person on bail is necessary and proportionate, and that an officer of the rank of inspector or above authorises the release on bail (PACE, 1984; PCA, 2017). ↩

4. Other safeguarding measures are available for specific types of cases. For example, child sexual exploitation cases tend to involve children’s services and civil protection measures, and alleged offences involving young people will often trigger the involvement of youth offending teams. ↩

5. Other than the established criminal law offences that serve to protect the integrity of the investigation, such as witness intimidation or harassment. ↩

6. The terms ‘domestic abuse’ and ‘domestic violence’ are used interchangeably throughout the literature. For the purposes of this review the term ‘domestic abuse’ will be used because it captures a broader range of behaviour which is now recognised as a criminal offence in England and Wales https://www.gov.uk/government/news/coerciveor-controlling-behaviour-now-a-crime [viewed 20/12/2020] ↩

7. The EMMIE framework looks at Effectiveness, Mechanisms, Moderators, Implementation and Economy. ↩

8. The Maryland Scientific Methods Scale (SMS) is a five-point scale ranging from 1 for evaluations based on simple cross-sectional correlations, to 5 for randomised control trials. ↩

9. The Police Powers and Procedures Statistical Bulletin states that “some of these forces mentioned quality concerns with their data, including partial returns. Therefore, data in this report give an indicative picture only, and should be treated with caution.” ↩

10. The ruling of the High Court in the Hookway case (Greater Manchester Police v (1) Hookway, (2) Salford Magistrates’ Court, AC, 19 May 2011) meant that suspects could be released on police bail for no more than 96 hours (four days). This ruling was quickly reversed, returning bail durations to an unlimited period. ↩

11. PCB, in terms of release from police detention, with or without conditions, was consolidated in the PACE Act (1984). An amendment to PACE was also introduced through the CJA 2003 (Criminal Justice Act, 2003) which gave police the power to attach conditions to bail for suspects who are released on bail without charge. This facilitated the CPS charging decision process. The CJA introduced the statutory charging scheme under which the decision to charge a person with a criminal offence (other than for minor offences) was to be made by a representative of the CPS rather than a police officer. In order to facilitate this, PACE amendments enabled the release on bail subject to conditions, pending the CPS’s decision, where a custody officer has decided there is sufficient evidence to charge a suspect (Cape, 2016). ↩

12. For court bail, it is only if a suspect fails to surrender to court on the designated date that constitutes a criminal offence. Breaches of other court bail conditions are not considered as criminal offences. ↩

13. The ratio of PCB to arrest was 37%, based on 27 forces. ↩

14. Martin’s (forthcoming) study of a single force found that PCB accounted for 19% of all initial disposals from police custody prior to the reforms. Phillips and Brown’s (1998) study also found that 17% of suspects were initially bailed for further enquires before a final decision was made about what action to take. Variations in force level figures are examined in more detail at the end of the section. ↩

15. Most analyses examine average duration per suspect but Martin (forthcoming) examines duration and other variables by recorded offence. ↩

16. In total, Martin found those offences where PCB was applied post-reform, 16% later switched to RUI in the course of the police investigation. These figures are based on two-years of data (April 2017– April 2019). ↩

17. The Home Office undertook a data collection exercise at one police station and manually collected data that covers two police force areas (Forces 1 and 2). One other police force also returned data (Force 3). 91 conditional PCB cases were reviewed over the week 07/10/2019 to 13/10/2019 (65% of conditional PCB cases in this week) in Forces 1 and 2. 58 conditional PCB cases were analysed from Force 3 between 07/10/2019 and 20/10/2019. ↩

18. The data from this request corresponds with when the PCB reforms were introduced in the PCA on 3 April 2017. Data are provided for one year immediately preceding the reforms (3 April 2016 to 2 April 2017) and one year immediately after the reforms (3 April 2017 to 2 April 2018). ↩

19. These data have been classified as experimental statistics meaning that they are in the testing phase and are not yet fully developed. Interpretation should be undertaken with caution. ↩

20. There is a two-day difference between the timeframes for the experimental PCB data collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by the Law Society (3 April to 2 April). ↩

21. There is a two-day difference between the timeframes for the arrest data collected for the Home Office Police Powers and Procedures publication (1 April to 31 March) and the FOI data provided by the Law Society (3 April to 2 April). ↩

22. Essex, South Wales, Suffolk, Surrey, Warwickshire and West Mercia. ↩

23. National data are available on police outcomes for recorded crime but do not classify whether a suspect had been placed on PCB or RUI. ↩

24. In the year ending March 2020, 54% of police recorded crimes in England and Wales with a domestic abuse flag received the outcome ’evidential difficulties -victim does not support’. This compared to 19% of non-domestic abuse offences. See ONS (2020). ↩

25. See Sleath and Smith (2017) for a useful UK empirical study of victim withdrawal in domestic abuse cases. ↩

26. Greater Manchester Police. ↩

27. Excluding GMP and including BTP. ↩

28. No data on the arrest to PCB ratio was available for the fifth force. Three forces recorded reductions in the volume of ‘victim does not support’ outcomes in year ending March 2018 with percentage reductions of between 5% and 25% but none of these forces were covered in the 28-force analysis in Figure 3. ↩

29. Postal requistions tell the suspect what they have been charged with and the date and time they should attend the firstcourt hearing (HMICFRS/HMCPSI, 2020). They report that it is common for suspects not to attend court as they mayhave never received the postal requisition, due to the police force having incorrect details or an out of date address. ↩

30. Although for bail the critical time is from arrest to charge or other outcome (crime outcomes data do not record date of arrest), these figures do at least provide a useful proxy for overall investigation duration. ↩

31. The Bail Process Project which was set up to improve the quantity and quality of information available to remand decision makers so they could make a more accurate assessment of the risk of offending on bail. ↩