Joint Committee On Human Rights Written Evidence



Dr David Best

Police Complaints Authority

19 September 2003


  From 2 October 2000, the Human Rights Act 1998 was incorporated into UK law, including Article 2 on the "right to life" stating that "Everyone's right to life shall be protected by law". The report below attempts to address issues around the right to life and its manifestation in police custody and broader contact between the police service and the general public, to examine the main causes of deaths in custody and underlying common factors in the context and characteristics of these deaths. One of the main emphases of the work will be on vulnerable populations, in the context of suicide, and in relation to ethnicity as an indicator of increased likelihood of custody death.

  According to Home Office data, just under 1.3 million people suspected of committing an offence are arrested every year. In 2001-02, of this 1.3 million arrests, 97,800 (8%) were recorded as being black, 55,600 (4%) as Asian and 11,800 (1%) as being of other "non-white" groups. This represented a rise of 7% for Asian and 12% for black arrestees when compared with the previous year.

  However, among all ethnic groups, death in custody is an exceptionally rare event, with the PCA Annual Report (2003) indicating that there were a total of 27 Category 3 custody deaths (see below for definition) in 2001-02 and 30 such custody deaths in 2002-03. Although the periods of assessment are not directly comparable, this means that there are slightly more than two deaths per 100,000 arrests in England and Wales.

  However, every such incident is a tragedy and is investigated accordingly. This investigation will be undertaken by the professional standards department of the police force in which the incident occurred or, on occasion, of an independent force. All of the deaths included in this analysis have also been supervised by an independent oversight body, the Police Complaints Authority (PCA).

  The PCA has previously attempted to address this issue in two linked reports (1999). The first of these highlighted risks around custody and made recommendations to address a number of risk factors with the follow-up report identifying successes—a success that is evidenced in the reduction in custody deaths over the course of the 1990s.

  The PCA is in a unique position to assess police-related deaths through its supervision role, by which all deaths involving the police are referred voluntarily by the relevant force for supervision of the investigation. The PCA file therefore is based on the police investigation and will supplement this with the relevant correspondence involved in the investigation of the death and with any matters arising.

  To make most use of the PCA evidence base, the paper will examine as many of the supervised cases involving the death of a member of the public as can be accessed in a five-year window, relating to the Home Office categories 1 to 3. The definition classes police-related deaths as:

    Category 1: Fatal road traffic incidents involving the police

    Category 2: Fatal shooting incidents involving the police

    Category 3: Deaths in or following custody

    Category 4: Deaths during or following other types of contact with the police

  For the purposes of the current analysis two groups have been excluded. Category Four deaths in custody and road traffic incidents involving emergency responses and standard patrol collisions. The reason for this is that each of these groups will include deaths among those not involved directly with the police. As the unit of analysis in the current study is the individual who has died, it was decided that the primary focus of the study would be on those who died in police custody, but to use an overview analysis of pursuits and fatal shootings as both a context and as contrast groups for the main analysis. In sum, what this means is that the results are presented as:

    1.  Summary characteristics of the 302 deaths considered in the report

    2.  Comparison of Category 3 custody deaths with pursuit and shootings deaths from the same period

    3.  Analysis of Category 3 custody deaths

    4.  Analysis of ethnicity factors in Category 3 custody deaths

  This, in effect, means that the study will consider fatal road traffic incidents, deaths in police care and custody and fatal police shootings that have been supervised by the PCA.

  The five-year window selected, from April 1998 to March 2003, is an attempt both to maximise the sample and to ensure that as much of the information that can be accessed within the PCA is used. There are limitations in data at both ends of the selection. For the earliest year, 1998-99, a number of the files have been destroyed and so some of these are likely to be excluded (in cases where no information can be accessed from other sources). For the most recent year, 2002-03, the problem is that many investigations have not been completed by the time of writing, and so there will be limited information available, and what information there is, will be of limited reliability as it will not have been confirmed through all the relevant investigative channels.

  The search strategy for identifying files was by using the PCA Annual Reports and then reconciling these with information accessed from the PCA computerised database or from the PCA Press Office files. Once the files had been identified, the physical tracking down of the case used one of four sequential mechanisms:

    1.  Was the file held by the PCA research department?

    2.  Was the file available in the storeroom of completed files within the PCA?

    3.  Was the file held at the PCA archives depot?

    4.  Is the file currently with the member or caseworker as it is an ongoing case?

  A total of 305 files were identified, but three of these files could not be tracked down, meaning that the total number of cases entered on the database is 302. For this reason, the data presented below, on the 302 files accessed, analysed and entered, is less likely to involve ongoing "live" cases (most likely from 2002-03) or those held in the archives (1998-99). Similarly, even among older cases, those that are particularly complex are more likely to be ongoing and so are less likely to be complete, for the purposes of the current analysis.


  The analysis outlined below is based on the 302 cases identified as falling into one of the first three categories, divided as below:

Fatal police shootings

  12  (4%)

Road traffic incidents

137  (45.4%)

Deaths in custody

153  (50.7%)

  This is reflected in the year breakdown, shown in Table 1 below

Table 1


1998-991999-2000 2000-012001-02 2002-03
46 (15.2%)57 (18.9%) 57 (18.9%)70 (23.2%)72 (23.8%)

  There are two reasons why this cannot be regarded as definitive trend data in spite of the apparent increase in numbers over the period of study. First, the availability of information relied on access to the file (although this was possible in almost all cases). However, it is also reliant on referral policy (particularly in road traffic incidents about whether the police involvement was sufficient to refer) and subsequent PCA decision-making about the acceptance of referrals. More importantly, the overall data disguise shifting patterns within the two main classes of death examined—fatal road traffic incidents and deaths in custody (as shown in Figure 1 below). Thus, while the total number of deaths in custody have remained broadly consistent over the course of the study, the number of fatal road traffic incidents have increased dramatically in the same period.

Characteristics of those who have died

  The mean age of those who died was 34.4 years, although it should be pointed out that there is a wide range with the youngest victim two years old (a victim in a road traffic incident) and the oldest 87 years old. The sample was also predominantly male (n=271, 89.7%), with only 31 deaths (10.3%) involving female victims. While all of the victims of police shootings were male, 20 of the female deaths were in road traffic incidents and 11 were deaths in police care or custody.

  Ethnicity data was available for 271 individuals (90%). Of these, 218 (80.1%) were classed as white, white British or white European, with 27 victims classed as black, black British or Afro-Caribbean (9.9%), 20 (7.4%) as Asian or Indian, and seven as from other ethnic backgrounds (2.6%).

  In terms of vulnerability, 90 of the 293 cases where this was known (30.6%), involved a victim with a previously identified mental health problem—in 26 cases this had been diagnosed by a psychiatrist, in 22 cases by the individual's GP and in 42 cases, mental health problems had been identified elsewhere (eg by prison doctors or social workers). While up to three separate diagnoses were recorded for six individuals, a history of self-harm was indicated in 29 individuals (32.2%), depression in 32 cases (35.6%), alcohol dependence in 21 cases (23.3%), drug dependence in 13 cases (14.4%) and schizophrenia or psychotic symptoms in 20 cases (22.2%). A further indication of substance use was derived from the post mortem toxicology reports, with an average of 1.2 active drugs reported from the toxicology reports. Of the 297 cases for which information was available, 178 individuals (59.9%) were positive for at least one drug or alcohol.

  Alcohol was the substance most frequently identified, present in 107 cases (35.4% of the total sample). With regard to illicit drugs, cannabis was identified post mortem in 50 deaths (16.8%), cocaine in 34 cases (11.4%) and heroin or heroin/morphine in 22 cases (7.4%). The other drugs frequently recorded were benzodiazepines in 38 cases (12.8%), although it is not clear if this was prescribed or illicitly diverted drugs and novel stimulants (such as ecstasy and its analogues) in 24 cases (8.1%).

  In terms of the location of the incident, large urban forces were, as would be expected, highly represented. A total of 48 cases (15.9%) occurred in the Metropolitan Police area, with the next highest prevalence in Greater Manchester (n=21, 7%), followed by West Yorkshire (n=17, 5.6%) then both Devon and Cornwall and Thames Valley (n=14, 4.6%). However, each of the 43 Home Office forces has experienced at least one death in the five-year time window (a full breakdown is presented in Appendix 1).

  Although difficult to classify, the reason the individual was initially in contact with the police was as a result of police intelligence in 166 cases (55.7%) or as a result of driving matters (89 cases, 29.9%). Other reasons for initial police involvement were being observed committing a crime (in 14 or 4.7% of cases), routine stops (in 11 or 3.7% of cases) or because the individual was reported to have been "behaving suspiciously" (in 13 or 4.4% of cases).

  Data were also gathered on the location of death although this can be misleading as it will often reflect where death was pronounced to be extinct rather than where the individual died. For this reason, the most common location was in hospital (in 130 cases, 43.5%), followed by in a public place (106 cases, 35.5%). Only 51 of the 298 cases on which this data was available (17.1%) actually died in police cells or other parts of the police station. A further 12 (4%) died at home.

  At the time of the incident, 124 of the 298 of the individuals for which the information was available (41.4%) were either detained or under arrest. In the remainder of cases, the police were attempting to arrest the individual (91 cases, 30.4%), there was no arrest intention (69 cases, 23.1%) or the death occurred after the individual had been released from police custody or contact (15 cases, 5%).

Causes of death and issues of concern for the police

  In 24 cases (8.1% of the 298 cases for which information was available) there was a restraint concern that was subject to investigation (although this concern was not necessarily upheld by the investigators), ten of these involving the use (and alleged misuse) of handcuffs. These include two cases where the individual was still in handcuffs when taken to hospital, one case where an individual was handcuffed while unconscious, and one occasion where the individual had flexi-cuffs used to restrain his or her legs. In four cases, the restraint issue relates to the use of CS spray, in one case to the type of restraining hold used by the police officer and in two cases to the length of time the individual was left sitting restrained in a police van. However, in a number of other incidents the matter investigated related to the technique used by officers to control the individual—including the use of a "bear hug" in one incident, forcing a man's arm up behind his back in another and the violent struggle preceding arrest in a third case.

  Similar problems exist in making sense of the cause of death identified at the post mortem, not least because in 13 cases, three causes of death are given while in a further 40 cases, two causes of death are provided by the pathologist. However, among the 301 cases for which at least one cause of death was available from the post mortem investigation, the most common main cause of death was "multiple injuries" in 76 cases (25.2%), followed by alcohol or drug toxicity in 47 cases (15.6%) and "head injury" in 33 cases (10.9%). Cause of death in all 12 fatal shooting cases was given as the effects of the shot or shots.

  In 10 cases (3.3%), "hanging" was given as the cause of death, with ligature strangulation given as the cause of death in one additional case. In four cases, excited delirium is given as the cause of death and, in five cases, asphyxiation is given as the cause of death. However, there are a total of 73 primary causes of death cited in cases included in the study.

Complaints, criminal and disciplinary outcomes

  In only 16 of 300 cases (5.3%) was there a formal complaint made by family members as a result of the incident that led to the death—10 of these arising in custody deaths, five relating to road traffic incidents and one following a police shooting. However, in several of these cases, more than one issue was raised in the course of the complaint (see Table 2 below). Furthermore, there are concerns expressed by family members that are recorded in the investigation file in a number of cases, while there may yet be complaints from family members in some of the cases that have not yet been completed.

Table 2

CategoryNature of complaint
PursuitOfficers knew the occupants and therefore should not have pursued; There was a delay in notifying the family about the incident; Witnesses stated that an officer kicked the deceased.
PursuitPassenger in pursued car complained that he was assaulted by an officer.
PursuitPursuing officer committed traffic offences in the pursuit; The pursuit was dangerous and should not have taken place; The FLO lied to the family; The officer was unsympathetic.
PursuitRelating to the actions of the police prior to the collision, at the scene, at the hospital and with regard to family liaison.
PursuitShould not have been pursued; conduct of the pursuit.
Death in custodyObvious ligature point in the cell.
Death in custodyUnlawful arrest; unnecessary force used.
Death in custodyOfficer failed to look after the individual's health and welfare; officer should have acted quicker; officer's actions contributed to the death; officer was aggressive and assaulted victim's mother.
Death in custodyFailed to provide proper medical attention; Custody officer listed victim as unknown when he was known to the police.
Death in custodyBreach of Code C: failure in duty.
Death in custodyUnlawful detention: officers acting outside the powers of the Mental Health Act; excessive use of force.
Death in custodyDeceased not allowed to use toilet nor to change clothes; FME pronounced the deceased fit to detain; Delay in informing family; FLO attitude not acceptable.
Death in custodyFailure of officers to make inquiries about drugs and alcohol consumption; Failure to update custody records; Failure to observe detainee properly.
Death in custodyInadequate investigation; Dishonesty of police officers; Failure of officers to intervene.
Death in custodyThe actions of officers caused the deceased to lapse into unconsciousness due to the failure to monitor prisoners.
ShootingInitial complaint of unlawful use of force and falsified accounts. Subsequent complaint alleged that: Family not informed soon enough about the death; FLO had no respect for family privacy; Offer to pay for funeral was made and withdrawn; Scene of shooting was managed insensitively.

  The final report records concerns expressed by family members or community groups in a further 12 cases where no formal complaint was made. These included:

    —  Concerns that the deceased should have been taken to hospital earlier.

    —  Concern that the time for arrest given by the police was inaccurate.

    —  Concern that the FME was not called, while the officers were watching football and that the officers did not do all they could to preserve safety.

    —  Concerns about the level of care and why the individual was not taken to hospital.

    —  Concerns about delays in informing the family and that clothing was seized.

    —  Dissatisfaction with the IO who was seen as rude by the family.

    —  Family concern that police actions contributed to the death.

    —  Concerns that the deceased was beaten and that potential witnesses were not requested.

  In only three of the finalised cases did officers face criminal charges for involvement in a death, and, on each occasion, was subsequently acquitted (although a further 31 cases have not yet reached this stage). These were:

    (a)  Five officers were acquitted of misfeasance and manslaughter following a death in a custody suite in Humberside.

    (b)  One officer was acquitted of misfeasance following a cell death in Lancashire.

    (c)  A police driver was found not guilty of dangerous driving following a police pursuit.

  In one further case, an officer was required to resign at a disciplinary hearing. In no finalised case were any officers demoted in rank following disciplinary proceedings.

  In terms of formal disciplinary proceedings, 16 of 260 cases (6.2%) resulted in officers receiving formal disciplinary outcomes. In eight cases an officer was admonished, in six cases this happened to two officers, and in one case each, three and four officers respectively were admonished or received a warning following investigations into police-related deaths. A total of 168 officers received advice from a senior officer in a total of 65 cases (of 260 that had reached disciplinary conclusions by the time of writing). In one case, a total of 15 officers were given advice following the investigation.

  At the time of analysis, information regarding recommendations was only available in 259 cases. Of these 259, policy, training or organisational recommendations were made in only 96 cases. The number of recommendations made in each case ranged between one and 30. In the remaining 163 cases (62.9%) no recommendations of any sort were made. Where recommendations had been made, information about implementation would not always be provided to the PCA, as the Authority has no role in monitoring the implementation of such recommendations.

  Finally, at the time of writing, an inquest had been held in 194 cases. In a further 56 cases the inquest date was yet to be set. In 45 cases no inquest was scheduled to take place, generally involving road traffic incidents where those driving the cars that killed the victims were later convicted of causing the death. Of the inquests held, 73 returned a verdict of "accidental death" (37.8%), with the jury in one further case returning a verdict of "accidental death contributed to by neglect". A further 26 cases (13.5%) resulted in verdicts of "misadventure" and 28 cases (14.5%) in verdicts of "natural causes". In 15 cases, an "open" verdict was returned, and in five cases the verdict was "lawful killing" (these were all police shootings). In a further six cases, the verdict was "drug-related death".

  Twelve cases (6%) resulted in "unlawful killing" verdicts (although all but one of these involved police pursuits and related to instances where the pursued driver had killed another road user or pedestrian). In only one custody case, relating to the death of a black man in Humberside, did a custody death result in a verdict of unlawful killing.

  Ten deaths resulted in verdicts of suicide and two in verdicts of "suicide contributed to by neglect". One of the 10 suicide deaths was a police shooting described by the coroner as a "suicide by cop" incident.

Comparison of death categories

  There was a significant disparity in age profiles with the mean age of deaths in custody (mean = 40.5 years) markedly higher than fatal shootings (mean = 33.7 years), which in turn was significantly higher than road traffic incidents (mean = 27.2 years; F = 28.1, p>0.001). The largest proportion of female deaths was in road traffic incidents (14.6%) compared with none of the 12 shootings and 7.2% of deaths in custody, which constitutes 11 cases (x2=5.73, p=0.06, ns).

  In terms of shifting patterns of deaths over the period of investigation, the basic trends are shown in Table 3 below:

Table 3

YearShootings (n=11) RTI (n=117)DIC (n=122)
1998-990 13   (9.5%) 33 (21.6%)
1999-20003 (25.0%)24 (17.5%) 30 (19.6%)
2000-012 (16.7%)26 (19.0%) 29 (19.0%)
2001-024 (33.3%)36 (26.3%) 30 (19.6%)
2002-033 (25.0%)38 (27.7%) 31 (20.3%)

  When ethnicity is compared by category of death, there is no significant difference. To enable this analysis, ethnicity categories were collapsed into four groups—white, Asian, black and other. For all three categories of death, over 75% of those who died were white. Of the 27 deaths classed as among black people, two (17.4%) were police shootings, 13 were in fatal road traffic incidents (48.1%) and 12 were deaths in custody (42.3%). Of the 20 individuals classed as Asian who died, 11 (55%) died in road traffic incidents and nine (45%) in deaths in custody.

  Those involved in road traffic incidents were less likely to come from identified vulnerable populations. With regard to a confirmed mental health indicator, this was the case for 9/12 (75%) of those fatally shot by the police, and 50% (n=75) who died in police custody, compared with only 4.5% of those whose status was known in road traffic incidents (x2 = 81.7, p<0.001). Similarly, while victims of fatal road traffic incidents averaged 0.7 different active drugs in their bloodstream at post mortem, the average for fatal shooting victims was 1.2 (in each case this included alcohol) and in deaths in custody the average was 1.6 (F = 16.6, p<0.001).

  There were significant differences in the disciplinary outcomes as a consequence of the category of incident (see Table 4)

Table 4

Shooting (n=11) RTI (n=117)DIC (n=122) F, sig
No of officers disciplined0.14, p  <0.01
No of officers given advice1.57 0.181.0611.61, p<0.001
No of policy recommendations4.6   0.4  1.3  13.46, p<0.001

  As is evident from the above table, road traffic incidents were markedly less likely than either of the other classes of police-related deaths to result in disciplinary outcomes or in recommendations of policy change by the SIO (which were most common in cases of police shootings). However, it is notable that the highest levels of both main disciplinary outcomes occurred in death in custody cases.

Focusing only on deaths in custody

  This section of the analysis will focus on the 153 cases of Category Three death according to the Home Office classification scheme. To re-iterate the main characteristics of the group, their mean age was 40.5 years (with a range of 15 to 76 years), and they were predominantly male (142/153 or 92.8%). A total of 27 individuals (17.6%) were categorised as non-white in the sample—nine as Asian, 12 as black and six as from other ethnic backgrounds. However, ethnicity data were missing on 11 cases (generally among files from the earlier years where the file was no longer available).

  On average, post mortem analysis revealed that they had consumed a mean of 1.6 active substances[64] in the period prior to their death—most commonly alcohol, which was detected in 67 cases (43.8% of cases). Other drugs consumed are shown in Table 5 below:

Table 5

Substance Number of cases %

  It is important to note that benzodiazepines will include those prescribed therapeutically—in some cases in custody—and so are not, unlike the other substances included, necessarily indicative of drug abuse.

  The group was rendered further vulnerable by the prevalence of mental health problems identified. Just over half of the cases for which information was available (75/149, 50.3%) had a prior indication of mental health problems—with 17 individuals having a previous diagnosis by a psychiatrist, 20 having GP indications of mental health problems and with the remaining 38 having other indications in the investigation files of earlier mental health problems. This is a level of mental health problems considerably in excess of that generally reported in custody populations (Bennett, 1998; Ingram and Johnson, 1998).

  For 30 individuals (19.6% of all custody deaths included) there were prior indications of anxiety or depression, 26 had recorded histories of self-harm (17%), 17 had markers for psychosis or schizophrenia (11.1%), 12 had histories of drug dependence (7.8%) and 18 (11.8%) had histories of alcohol dependence. "Behaviour problems" or other psychiatric problems were recorded in six further individuals. In other words, for a substantial proportion of the custody death group considered in this report, there were not only indications of mental health problems and/or substance abuse, but there were previous contacts with health agencies attempting to address these problems.

Locations of the death

  Again, the police service most commonly associated with custody deaths was the Metropolitan Police Service (MPS) accounting for 32 (20.9%) cases. The force with the next highest level of custody deaths was Northumbria (n=10), followed by West Midlands and Devon and Cornwall (eight deaths each). However, the 153 deaths were spread between 36 different police forces in England and Wales.

  In more specific terms, the place of death was recorded as the police station or cell in 45 cases and "police vehicle" in a further six cases—in other words 33.8% of the 151 cases for which this information was available involved death in a police location. The other main locations for death recording were in hospital (79 cases or 52.3% of the valid sample), with 12 individuals dying in a public place and nine at home.

  The initial contact resulted from "police intelligence" in the vast majority of cases (111 or 74.5% of the death in custody cases). Less frequent reasons for the initial police involvement were traffic or driving matters (in 13 cases), the individual being observed committing a crime (12 cases), the police perception that the individual was engaging in suspicious behaviour (seven cases) or routine stops (four cases). Other reasons were given in a further two cases and this information was not available in four cases. As has been detailed above for all deaths, there were restraint-related aspects of the investigation in 23 cases (15.4%)—generally relating to the timing and location of handcuffing, the use of force in the initial arrest attempt, the use of CS spray or delays in the removal of restraints when it was apparent that the individual was experiencing significant health problems.

  Although custody detention issues are likely to be significant, the attempt to quantify this is problematic. The calculated mean time is 384 minutes (just over six hours), but this is heavily skewed by two cases where the individual is in custody for more than two days. However, it is worth noting that in nine cases, the individual is in detention for 24 hours or more.

  The prevalence of substance use is clearly indicated by the fact that "toxicity" is cited as a cause of death in 47 cases (31.8% of the 148 cases for which this information is available). Head injuries are cited in 12 cases, hanging in 10 cases, multiple injuries in seven, hypoxia in five cases and excited delirium in four cases. In at least seven further cases, alcohol-related factors are cited among the causes of death. Changes over time in the frequency of both toxicity deaths and hanging deaths are given in Figure 2 below:

  As can be seen from the above table, there is no consistent pattern in deaths where toxicity is given as a cause, but there appears to be a positive change in the number of strangulation or cell deaths with only one death in each category in each of the last two years.

Investigations and outcomes

  In only two of the 131 cases that have reached that stage was there a criminal trial of an officer, with six officers tried in total, with all six being acquitted. An officer was required to resign following one of the custody deaths, and in 14 cases a total of 25 officers were warned or admonished following the investigation. In 45 of the 127 cases (35.4%) completed to date, officers were given advice by a senior officer, resulting in a total of 134 officers dealt with in this way.

  In 49 of these 127 cases (38.6%), the Senior Investigating Officer (SIO) recommended further action at a policy level. The most common areas identified by senior investigators as needing to be addressed were:

    —  Training needs for officers (n=39, 30.7%).

    —  Issues around the management of vulnerable populations (n=31, 24.4%).

    —  Equipment issues (n=30, 23.6%).

    —  Incident management issues (n=21, 16.5%).

  The other most common themes identified by SIOs were custody management issues (in nine cases), failures of inter-agency working (in seven cases) and failures of post-incident investigation (in five cases).

  A further investigation was carried out in the form of a coroner's inquest in 112 cases. Details of the inquest outcomes are given in Table 6 below. It should be noted that multiple verdicts are given in a number of cases:

Table 6

Inquest verdictFrequency %
Accidental death2919.1
Drug related63.9
Accidental death contributed to by neglect 10.7
Natural causes2818.4
Suicide by neglect2 1.3
Accidental death contributed to by restraint and failure
to provide medical care
Misadventure contributed to by police neglect 10.7

  For the 10 complaints received to date (details of which are provided above), four are still under active investigation. In the remaining cases, one has been informally resolved, two have been upheld in part and three have not been upheld.

Examining ethnicity issues among custody deaths

  This part of the analysis will focus on the 27 non-white individuals who died in custody in the period of the review. The mean age of the non-white group was 37.3 years (range of 19-66 years) and consisted of 26 males and one female.

  Seven of these individuals (26.9%) had a previous indication of mental health problem(s)—three on the basis of psychiatrist diagnosis and four based on information from other sources in the SIO's report (information was missing in one case). Four of these individuals had drug dependence indicators and one had an alcohol dependence marker. Two individuals had indications of schizophrenia, two had markers for either anxiety or depression, and one had another psychiatric problem. In other words, there were a total of 10 symptoms indicated in this group.

  In 22 of the 26 cases where this information was available (84%), there was at least one active substance present at the toxicology. As with the larger sample, the most common substance present was alcohol (48%), followed by cocaine (28%), heroin or morphine (20%), cannabis (20%) and benzodiazepines (16%). There was no relationship between mental health status and the likelihood of substance use prior to death.

  More than half of the cases (n=14, 53.8%) of custody deaths involving ethnic minority individuals occurred within the MPS, with a further four deaths occurring in West Midlands, two in Sussex, and one each in Northumbria, Essex, West Yorkshire, Hampshire, Surrey and Hertfordshire.

  For the ethnic minority group, just over half the deaths occurred in hospital (13/25, 52%) with four occurring in police stations or cells and one in a police vehicle. Three individuals died at home and four in a public place. In the majority of cases (17/25), the reason for the initial police contact was based on police intelligence, and also for the majority (18/26) the individual was under arrest at the time of the death. However, for a further five, death occurred post release, and for the final three individuals death occurred either while the police were in the process of attempting to arrest or detain the person.

  In five of the 23 cases for which this information was available, there was a restraint issue, four relating to the method of restraint (generally about the use of handcuffs) and one involving a violent struggle prior to the arrest.

  However, in 11 cases (42.3%) toxicity was cited as a cause of death, with multiple injuries cited in three cases, and excited delirium and head injury in two cases each. The full range of primary causes of death are given in Table 7.

Table 7

Post mortem cause of death
YearFirst SecondThird Inquest verdict
1998-99Toxicity Unknown
Advanced alcoholic liver disease Natural causes
Multiple injuries Accidental death
Hypothermia Misadventure
Toxicity Accidental death
Dilated cardiomyopathy Natural causes
Toxicity Misadventure
BronchopneumoniaInhalation of vomit ToxicityAccidental death
Excited delirium Pending
Not known Unlawful killing
1999-2000Toxicity Accidental death
Irreversible cerebral anoxia ToxicityMisadventure
Ischaemic heart disease Coronary heart diseaseToxicity Natural causes
Head injury Accidental death
HepatitisMulti organ failure Open
Head injury Accidental death
Excited deliriumToxicity Drug related
Asphyxiation Misadventure
2000-01Toxicity Accidental death
Unknown Open
2001-02Toxicity Misadventure
2002-03Multiple injuries Pending
Chronic bronchitisEmphysema Pending
Cardiac arrest Pending
Respiratory distress syndrome ToxicityPending
TracheobronchitisSkull fracture Pending
Multiple injuries Accidental death

  In outcome terms, there has been one trial in the 22 completed cases, with 21 inquests having taken place. The details of inquest verdicts are given in Table 8 below:

Table 8

Inquest verdictFrequency %
Accidental death829.6
Drug related13.7
Natural causes311.1
Unlawful killing13.7

  In four of the 22 cases completed to date, a total of eight officers were warned or admonished and a further 21 officers were given advice by senior officers.

  When inferential statistical comparisons were carried out comparing white and non-white custody deaths, almost no statistical differences emerged. Although white custody deaths were typically older (40.7 years versus 37.3 years) this was not significant.

  A significantly lower proportion of non-white deaths in custody involved mental health problems (28% versus 55.4%; c2 = 6.12, p<0.05).

  There has also been a significant reduction in the proportion of non-white deaths over time within the window of investigation in the study, which does achieve statistical significance.

  As can be seen from the figure above, the number of deaths among non-white individuals decreased over the first four years of investigation but has increased in the most recent year, while the total number of white deaths has remained relatively constant over the period of investigation.

  Finally although it did not attain statistical significance and the numbers are relatively low, it is notable that there are restraint issues in a higher proportion of the deaths involving non-white individuals (21.7%) than among white individuals (12.3%) (see Appendix 2).

  In terms of investigation outcomes, a slightly higher mean number of officers received warnings or admonishments in non-white deaths (mean = 0.36) than in white deaths (mean = 0.18) but this was not significant. Conversely, more officers received advice on average for white custody deaths (mean = 1.2) than in non-white custody deaths (mean = 0.9) but this also failed to attain significance.


  The study shows a relatively consistent pattern of custody deaths in the five-year window examined, with the 153 PCA-supervised custody deaths roughly equally spread across the period of investigation. This contrasts with the comparison group of pursuit deaths where the trend is upwards over the period of assessment, with the exception of the most recent year. Although exceptionally low as a proportion of arrests (153 deaths from around 6-7 million arrests in the period of investigation), it would not indicate that recent developments in monitoring or training are having a resulting effect on the overall number of fatalities. However, it is important to note that, over the last 10 years, the trend has been towards reduced numbers of deaths, particularly from hangings, the cause of death that is most obviously preventable. However, similar improvements have not been detected in the prevention of alcohol and drug-related deaths.

  Within this custody group, there are marked variations in the demographic characteristics of those who have died. The group are primarily male, more than 80% are of white ethnic origin, and with a mean age of around 40 years (although there are a broad range of ages). There is an over-representation of ethnic minority individuals in custody deaths in the five-year window studied—while 17.6% of deaths are of individuals classed as non-white, the 2001 census for England and Wales reported that 9% of the population are from BME groups. It is also slightly higher than the arrested population reported by the Home Office for 2001-02, which showed that 13% of the total arrested population were from minority groups. While this may partly relate to the way ethnicity has been classified in some cases (and inconsistencies across measures), the over-representation of ethnic minority groups among the deceased group should not be ignored.

  One of the main findings of the study is the exceptionally high prevalence of mental health problems recorded in the police investigation reports, at around 50% although higher among white deaths than among ethnic minority groups. In contrast, deaths among minority detainees were slightly more likely to result in the investigation considering aspects of the restraint of the detainee. However, there are very few clear, statistically identifiable group differences in the characteristics of the incident or the individual as a function of ethnicity, although the small numbers of individuals from BME groups means that statistically robust differences are difficult to detect.

  The general issue of vulnerability cannot be overstated. The preponderance of alcohol consumption and illicit drug use (particularly relating to the use of both cocaine and benzodiazepines) is a risk in its own right and compounds the risk associated with mental health problems. The latter, referred to as dual diagnosis (Strathdee et al, 2003), is associated with markedly increased risk of mortality from both custody deaths and from police use of firearms.

  One of the most contentious issues will be the apparently low levels of culpability for police officers resulting from the total of 300 cases. Although seven officers were charged with criminal offences, none were convicted. One officer was required to resign and none were demoted. Similarly, in only one case, did an inquest verdict of "unlawful killing" relate to police activity (in the same case that five officers were charged and subsequently acquitted of criminal offences). However, in many of these cases, none of the adjudicating bodies—the inquest, the Crown Prosecution Service, the police investigators or the PCA supervising members—have found fault with police conduct and many of the disciplinary outcomes relate to ancillary matters rather than the actual cause of death.

  Before drawing tentative conclusions, it is important to acknowledge the limitations of the study. All of the analyses are based on summaries of the PCA file, which in turn is heavily reliant on the final police report into each incident. These reports not only vary in depth and quality, they are also designed for a purpose other than research and so may not be consistent with the aims of the project.

  However, there are a number of key inferences that can be derived. Although there have been significant gains, it is essential that the police remain vigilant and seek to eliminate the preventable deaths that do, on occasion, still occur. However, deaths are not randomly distributed across the population or indeed the arrest population, and this is mediated by incident and response type. Vulnerable populations (those with a mental illness and/or users of alcohol or illicit drugs are hugely over-represented), while those from ethnic minorities are less likely to have a recorded mental health problem but are likely to be slightly younger and slightly more likely to have been involved in an incident that provoked concerns about the method of restraints.

  There are a number of implications of this for training and supervision. Earlier access to medical interventions are essential as is first aid training and refresher courses for all officers involved in custody. Similarly, officers must be made aware of the risk factors for self-harm and mental health problems, and for ensuring that a "safety first" approach is adopted in custody suites. It may appear that, following the significant gains in the late 1990s, some forces may have allowed complacency to creep in thus generating risks for those held in custody.

  All deaths in custody are, at one level, preventable, although in practice, this is obviously not achievable as many "natural causes" deaths may be completely unrelated to any actions on the part of the police. The repetition of areas of recommendations from final reports suggests that prevention is not yet a sufficient objective and that some opportunities for organisational learning are not being taken. To ensure that HRA requirements are adhered to, the police service must ensure that lessons are learned and that deaths, particularly those involving vulnerable groups, are minimised.

  All analyses are also made more problematic by the huge variations in the causes of death identified at post mortem, and this is reflected in the inquest verdicts passed down. In only one cases was the inquest verdict "unlawful killing" in relation to actions of the police and, given this finding, it is perhaps not surprising that only one officer was charged with a criminal offence. The most common disciplinary outcome (in cases where this arose was either formal admonishment or advice for officers), and the recommendation of organisational issues varied markedly from case to case, although much more common in shootings cases than in deaths in custody or even more markedly when compared to road traffic incidents.


Frequency %
Valid Percent
Cleveland31.0 1.01.0
Devon & Cornwall14
Northumbria124.0 4.09.6
Metropolitan Police48 15.915.925.6
Essex72.3 2.327.9
Leicestershire31.0 1.028.9
South Yorkshire51.7 1.730.6
Merseyside134.3 4.334.9
Greater Manchester21
Durham31.0 1.042.9
South Wales113.7 3.746.5
Lincolnshire51.7 1.748.2
Sussex82.7 2.750.8
West Midlands124.0 4.054.8
West Yorkshire175.6 5.660.5
Avon & Somerset7 2.32.362.8
Dorset41.3 1.364.1
Gloucestershire31.0 1.065.1
Hampshire72.3 2.367.4
City of London1.3 .367.8
Derbyshire31.0 1.068.8
Gwent2.7 .769.4
Surrey72.3 2.371.8
Kent1.3 .372.1
Bedfordshire31.0 1.073.1
Nottinghamshire31.0 1.074.1
Thames Valley144.7 4.778.7
North Wales51.7 1.780.4
West Mercia41.3 1.381.7
Staffordshire72.3 2.384.1
Wiltshire1.3 .384.4
Norfolk41.3 1.385.7
North Yorkshire41.3 1.387.0
Cheshire31.0 1.088.0
Humberside41.3 1.389.4
Hertfordshire1.3 .389.7
Cambridge2.7 .790.4
Lancashire93.0 3.093.4
South Yorkshire31.0 1.094.4
Dyfed Powys51.7 1.796.0
Cumbria31.0 1.097.0
Warwickshire31.0 1.098.0
Northamptonshire31.0 1.099.0
Suffolk2.7 .799.7
British Transport1.3 .3100.0
Total301100.0 100.0


EthnicityWhat was the restraint issue Cause of death Coroners verdict
whiteUse of CS on person with MH problem Organ failureSuicide
whiteHeld around the chest in reverse bear hug. Brain injuriesAccidental contributed by restraint and failure to provide
blackComplaint about handcuffing Excited deliriumpending
blackdragged from van—then left on ground with handcuffs on Multiple Injuriesunlawful killing
blackhandcuff marks, but pathologist could not comment on role restraint played in death Toxicityaccidental death
white Excited deliriumaccidental death
whitedeceased had spent over half an hour sitting handcuffed in the van outside the custody suite Aspiration of stomach contentsmisadventure
whiteuse of CS Toxicitymisadventure
whitewas handcuffed during hospital transfer HypoxiaAccidental death
asianofficers use of flexi-cuffs to restrain legs in contravention of force police Toxicityaccidental death
deceased was conveyed to hospital still cuffed Myocardical infarctionpending
restrained for hospital transfer Multiple Injuries
otherviolent struggle prior to arrest Unknownopen
whitehandcuffed white unconscious Toxicityunknown
whiterestrained at hospital due to fear of assault on staff—handcuffs not suitable for long term use Acute alcohol withdrawal with ketosis
whitepolice took hold of man's arm, put it up his back and forced him to the ground Ischaemic heart diseasepending
black Cardiac arrestpending
whitepolice assaulted him during arrest but no evidence to link to stroke Bronchopneumoniapending
white Excited deliriumpending
white Cardio-respiratory failurepending
white Unknownpending
.blood in cell. no explanation as yet Inconclusivepending
whitestruggled to put on handcuffs and CS used twice Inconclusivepending

64   Active substances refers to illicit drugs, alcohol or prescribed medications (including those diverted through illicit routes). Back

previous page contents next page

House of Lords home page Parliament home page House of Commons home page search page enquiries index

© Parliamentary copyright 2004
Prepared 26 January 2004