Manual restraint of adult

psychiatric inpatients: a
literature review
Report from the Conflict and Containment
Reduction Research Programme
Duncan Stewart, City University
Len Bowers, City University
Alan Simpson, City University
Carl Ryan, City University
Maria Tziggili, City University

June 2009
Department of Mental Health and
Learning Disability
City University
London E1 2EA

Levels of patient violence on psychiatric wards are high. It has been estimated that
almost half of nursing staff and one in seven patients are subject to a physical assaults
per year (Healthcare Commission 2007). Although the majority of such attacks result
in little or no physical injury, psychological responses can be significant, with reports
of consequent anger, anxiety, post-traumatic stress disorder symptoms, guilt, selfblame and shame (Needham et al. 2005). Manual restraint is a method of last resort
for the management of violent or challenging behaviour, not least because of concerns
expressed about the safety of patients under restraint (Blofeld et al. 2003;Paterson et
al. 2003). Training courses for nursing staff on the prevention and management of
violence and aggression, often referred to as Control and Restraint, are now
mandatory for UK psychiatric service providers (National Institute for Mental Health
in England 2004). However, there is a notable absence of any controlled studies to
support the effectiveness of manual restraint techniques (Sailas & Fenton 2005).
In this review, manual restraint is defined as physically holding the patient to prevent
or restrict movement. This is distinct from physical contact during the process of
putting patients into mechanical restraints. As a consequence, much of the literature
included in the review comes from the UK where mechanical restraint is rarely used.
Despite a large literature on the nature of violent behaviour among psychiatric
inpatient populations, there is little published data on the frequency of use of manual
restraint in the UK, or the antecedents and consequences of manual restraint use
beyond the immediate act of aggression or violence itself.
Literature search
This review was conducted in parallel with a review of mechanical restraint
techniques. Electronic searches of the main databases were conducted to locate post1960 empirical studies of restraint in English. The databases searched were: PsycInfo,
Cochrane, Medline, EMBASE Psychiatry, CINAHL and the British Nursing Index.
Key words utilised were restrain$, psych$ and mental$. Consistent with the aims of
the review, the following thesaurus terms were excluded: child, eating disorder, diet,
dementia and elderly. Resulting titles and abstracts were then inspected for relevance.
Evaluations of aggression management techniques were included if they specifically
concerned manual restraint, but studies of breakaway or self-defence techniques were
excluded. The type of restraint (manual or mechanical) was not always apparent from
the abstracts: where there was any ambiguity, the original was obtained an inspected.
As the literature accumulated, further references were obtained by following up
citations. The final number of identified empirical studies of manual restraint was 43.
All but four studies were from the UK. The others were from Canada, Australia (two)
and New Zealand.
Methodologies of the studies reviewed
Half (n=22) of the studies were retrospective analyses of official incident records,
although this was sometimes supplemented with descriptive data from other
sources. Nine studies used questionnaires or devised non-routine incident forms to
collect data from nurses on aggressive incidents on wards and outcomes in terms of
the use of restraint and other forms of containment. Four studies used a repeated
measures design to measure the effectiveness of various aggression management
training programmes. However, these assess the impact of training only in terms of

staff confidence and acquisition of skills to undertake restraint and do not measure the
use of restraint pre and post the programme. Finally, the review includes five
qualitative studies which describe staff and patient experiences of manual restraint
and three case studies of death during restraint. The studies were conducted in
various types of ward, ranging from acute wards (n=16), secure units (n=9), both
acute wards and secure units (n=4), general adult wards (n=2), to a mix of wards
(including several categories; n=9). Three studies did not specify the type of ward.
Given the diversity of settings, it is likely that patient populations varied greatly
between studies.
Analytic procedure
The aim of the literature review was to establish existing evidence for and against the
working model and assess commonality and links between different conflict and
containment types such as patient profiles, chains of events, patient experiences,
circumstances of use, etc. A structured data extraction tool was created with various
headings including sample, methodology, admission status, age, gender, ethnicity,
ward type, service setting, risk status, time spent on ward, rates of restraint,
antecedents/causes, patients views, staff views, etc. Where published papers
provided empirical evidence, this was entered on the tool. The headings of the
resultant matrix have then been summarised for the purposes of this review. A
hierarchy of evidence was established to rate the weight of each study in relation to
the projects aims. The most weight was given to studies conducted in the UK, on
acute wards and/or PICUs, studies with large samples and to findings replicated
across studies.
Incidence
As most studies were small scale, local (based in specific hospitals) and were few in
number, it is hard to draw any firm conclusions about the general extent of the use of
restraint. Twenty-one studies reported the incidence of manual restraint: either as the
number of episodes over a given time period (n=10) or as the proportion of responses
to violent incidents on the ward (n=11).
For studies reporting the total number of restraint incidents, rates per 100 beds have
been calculated where possible to aid comparisons. The rate of restraint use varied
from: 9.3 per 100 beds per month (Ryan & Bowers 2006), 12.9 per 100 beds per
month (Southcott & et al. 2002), 17.9 per 100 beds per month (Leggett & Silvester
2003) to 25.8 per 100 beds per month (Parkes 1996). Unadjusted rates showed much
greater variation, illustrating the utility of presenting rates of restraint and other events
on the wards in standardised form. It was not possible to calculate a standardised rate
for five studies. The first reported an average of 18.9 restraint episodes per month
across a mental health trust over a three year period (Lancaster et al. 2008). A survey
of nurses asked respondents to estimate the frequency of restraint use on their wards,
providing an average of 3.1 per month (Wright et al. 2005). A 2005 census of
inpatients in England and Wales found that 8% had at least one episode of control and
restraint during their current stay (Healthcare Commission 2005). Subsequent reports
in 2006, 2007 and 2008 showed rates of 8%, 11% and 12% respectively. A higher

rate (18%) was reported by a study of 12 acute wards, although the sample was
restricted to patients admitted for at least two weeks (Bowers et al. 2003).

Other studies have used violent or aggressive incident reports as the basis for analysis,
which may not include all restraint episodes (see antecedents and circumstances
section below). Nevertheless, the proportion of violent incidents which resulted in
restraint varied markedly, with some studies in forensic settings reporting very high
level of restraint. Rates varied from: 12% (Torpy & Hall 1993), 22% (Parkes 2003),
23% (Southcott & Howard 2007b), 26% (Tobin, Lim, & Falkowshi 1991), 36%
(Duff, Gray, & Brostor 1996); Kennedy et al 1995, 38% (Shepherd & Lavender
1999), 57% (Dowson, Butler, & Williams 1999) to 67% in a medium secure unit
(Gudjonsson, Rabe-Hesketh, & Wilson 2000) and 76% in a special hospital for
dangerous offenders (Larkin, Silvester, & Jones 1988).
One important issue is the extent to which patients are repeatedly restrained. Few
studies recorded repeated restraint episodes which occurred for the same patients,
usually because the main focus of the research was the incidence of aggressive or
violent behaviour rather than its management. In only two cases was it possible to
extract or calculate a mean rate of restraint use per patient. Lancaster et al.s (2008)
study of 680 incidents of physical restraint reported the average number of restraints
per restrained patient as 2.6 (with a maximum of 33 for a single patient). Data from
Leggett and Silvester (2003) suggest a mean of 4.9 separate incidents of restraint per
restrained patient. Another study reported that 15% of restrained patients had been
subject to more than one episode of restraint (Smith AD & Humphreys M 1997). The
lack of information on repeatedly restrained patients makes it difficult to interpret the
prevalence of manual restraint as there may be particular sub-groups of patients
whose behaviour makes them more likely to be restrained.
Information on the duration of restraint episodes was also sparse: only three studies
reported an average duration of restraint. A study of 557 restraint incident forms over
a four year period showed an average duration of 12 minutes (Leggett & Silvester
2003). Riley et al. (2006) compared the average duration of restraint in the supine
(patient on the floor on their back) and prone condition (patient on the floor on their
front) and found both to be 10 minutes, whilst a comparison of horizontal and vertical
(standing, sitting or kneeling) positions found a mean of 10 minutes for the former
and 5 minutes for the latter (Whittington et al. 2006).
Restraint techniques
Three papers using the same data-source specifically examined the use of different
restraint positions (Lancaster, Whittington, Lane, & Riley 2008;Riley et al.
2006;Whittington, Lancaster, Meehan, Lane, & Riley 2006). Of the 680 reported
incidents of restraint, the majority (58%) were in the horizontal position (Lancaster,
Whittington, Lane, & Riley 2008). When analysis was restricted to the first restraint
episode for each patient (n=261) the proportion restrained in the horizontal position
was slightly higher (62%). Among incidents involving horizontal restraint, the
majority were in the prone position as apposed to the supine position (56% vs 44%;

(Lancaster, Whittington, Lane, & Riley 2008). The proportions were similar when
analysis was restricted to the first restraint episode (58% vs 42%; (Riley, Meehan,
Whittington, Lancaster, & Lane 2006).
Two national surveys asked psychiatric nurses which restraint techniques they use
most frequently. The first (Lee et al. 2001) found that restraining holds the threeperson teams were as frequently used as verbal de-escalation (reflecting the
prominence of these techniques in training). Wright et al., (2005) found that 31% of
nurses reported that the restraining hold was used on their ward, 24% the three-person
team and 21% taking the patient to the floor in the prone position. The difference
between the two studies in the proportion of nurses using these restraint techniques
may reflect the way the questions used in the survey. Lee et al. (2001) explicitly
asked respondents about the techniques they personally use, while Wright et al.
(2005) asked about the techniques used on respondents' wards. Differences may also
reflect the nature of the survey samples. Lee et al.'s study was confined to staff from
PICUs and Regional Secure Units, whilst Wright et al. sampled from all acute
inpatient psychiatric services. Another study found that high levels of staff
satisfaction with their ability to put restraint techniques into practice (Southcott &
Howard 2007a). When patients are taken down to the floor nurses reported that this
was in a controlled manor in the majority (84%) of cases and that physical holds were
usually successfully established and maintained (93% of cases).
Antecedents and circumstances of manual restraint
Eight studies reported the antecedents of restraint use. As might be expected, a
commonly cited reason for restraint was violent or aggressive behaviour. However,
these studies consistently identified a variety of antecedents to restraint including
attempts to abscond (Bowers et al. 2003; Gudjonsson et al 2004; Ryan and Bowers
2006; Smith and Humphreys 1997; Southcott et al., 2002), disruptive behaviour
(Ryan and Bowers 2006; Smith and Humphreys 1997), agitation (Gudjonsson et al
2004), verbal assault/threat (Duff et al 1996; Southcott et al., 2002; Smith and
Humphreys 1997), refusal of medication (Bowers et al. 2003; Gudjonsson et al 2004;
Ryan and Bowers 2006; Southcott et al., 2002), self-harm (Smith and Humphreys
1997; Southcott et al., 2002), and property damage (Ryan and Bowers 2006; Smith
and Humphreys 1997). A study of restraint request forms found that violence was
rarely mentioned as a cause for restraint; more general challenging behaviour was the
most frequently cited reason (Ryan and Bowers, 2006). Over half (52%) of restraint
requests were classified as an emergency response (e.g. to an attempted abscond)
while 48% were classified as planned following patients refusal to comply with
instructions. One study reported that restraint was initiated following assaults on staff
(29% of cases), but other reasons were not reported (Parkes, 1996) while another
found restraint (as opposed to counselling or medication) to be more likely when
some form of staff-patient interaction was involved (Tobin et al 1991). Qualitative
interviews with staff have identified a poor ward atmosphere and failed
communication between staff and patients as antecedents for restraint (Bonner et al
2002).
The most rigorous paper involved a multivariate analysis of 1,515 untoward incident
forms (excluding self-harm and suicide attempts) on general psychiatric wards

recorded over a three year period (Gudjonsson et al., 2004). The most frequent
antecedents recorded on the forms were agitation (32%), specific interaction with
patients or staff (24%), staff refusal of patient's request or patient's refusal to take
medication (16%) and attempted abscond (7%). When other variables (e.g. patient
characteristics) were controlled for, the use of manual restraint to manage incidents
was predicted by attempts to abscond, staff denying a request and the patient being
rated as agitated. A nurse being the target of assault was also associated with an
increased likelihood of restraint.
It is interesting that some of the least frequent antecedents identified on incident
forms were the strongest predictors of restraint in the statistical model. This suggests
that studies which simply provide a descriptive account of officially recorded reasons
for incidents need to be interpreted with caution. Causes identified on incident forms
are not necessarily the same factors which determine the management approach to an
incident. Gudjonsson et al (2000, 2004) and Shepherd and Lavender (1999) found
that incidents rated as being more serious (usually involving violence or injury)
increased the probability of restraint. There is also evidence that restraint is more
likely to be used if the target of violence is a staff member as opposed to other
patients, self or property (Tobin et al 1991; Parkes 2003).
The decision to manage an untoward incident with restraint can be influenced by
aspects of staff-patient interaction which are not usually recorded. For example, some
studies suggest that the type of restraint used varies by the way nurses perceive the
cause of an incident. Patients refusal to communicate has been associated with the
use of the supine position and the threat of imminent violence with the prone position
(Riley et al., 2006). Another study used multivariate techniques to identify factors
associated with restraining patients on the floor (versus standing or sitting;
Whittington et al., 2006). This position was more likely if patients were formally
detained, had self harmed, had unclear thoughts prior to an incident and increased
voice volume, but was negatively associated with age, perceived causes such as bad
news and personal gain, and where there were no obvious warning signs before the
incident.
Outcomes of manual restraint
Two main outcomes are identified in the review: injuries (to staff or patients) and use
of other containment methods.
Injuries
Manual restraint carries a risk of injury for both patients and staff. In the most severe
cases this has involved the death of patients. Two case studies (Morrison and Sadler,
2001; Patterson and Leadbetter, 1998) describe the circumstances of deaths resulting
from positional asphyxia during manual restraint. Prolonged use of restraint in the
prone position was also implicated in the David Bennett inquiry (Blofeld et al 2003).
Patterson et al (2003) identified 12 cases of patient deaths during restraint between
1979 and 2000 across a range of UK health and social care settings and concluded
that downwards pressure on the chest to hold a patient in the prone position should be
avoided. The study also highlights other factors such as physical conditions,
substance use and prescription of neuroleptic drugs which may heighten the risks
associated with restraint use for some patients.

Non-fatal injuries during restraint tend to be more common among staff than patients.
The proportion of restraint episodes resulting in staff injuries ranged from 12%
(Southcott and Howard, 2007); 17% (Riley et al., 2006; Lancaster et al, 2008); 19%
(Parkes, 1996; Leggett & Silvester, 2003; Parkes, 2003); to 40% (Dowson et al.,
1999). The proportion of restraint episodes resulting in injuries to patients ranged
from 5% (Southcott and Howard, 2007); 6% (Riley et al., 2006; Lancaster et al.,
2008); 7% (Dowson et al., 1999); 10% (Parkes, 1996); to 18% (Leggett & Silvester,
2003). One survey of nurses found that 13% reported patient injuries during the last
occasion they were involved in the use of restraint, but 22% reported staff injuries and
that injuries sustained by staff were generally more serious that those sustained by
patients (Lee et al 2003). Indeed, Harris and Rice (1986) found that staff lost more
days on average for injuries that occurred during restraints than for injuries that
occurred during assaults. The likelihood of injuries occurring may also depend upon
the nature of the incident preceding the use of restraint. A multivariate analysis of
680 restraint episodes showed the risk of staff injury was increased when an assault
had taken place, but patient injuries were more likely if the patient had self-harmed,
used substances or used a weapon prior to restraint (Lancaster et al, 2008).
Despite the risk of death associated with restraining patients face down, from the
evidence available to this review there appears to be no significant difference in the
prevalence of staff or patient injuries by whether patients are restrained on the floor in
the supine or prone position (Riley et al., 2006). However, there is a slightly greater
risk of staff injury when restraining a patient in the prone position compared to the
standing position (Lancaster et al., 2008).
Other containment methods
Restraint was often followed by the use of other containment methods, usually
medication and less frequently by seclusion. Ryan and Bowers (2006) reported that
half (51%) of nurses requests to restrain patients resulted in the patient receiving
medication and 17% resulted in seclusion, although the study does not report rates for
actual restraint episodes. A study of 2,180 violent incident forms found that
restrained patients were sedated on 44% of occasions compared to a fifth of cases
where restraint was not used (Gudjonsson et al., 2000). Similarly, Shepherd and
Lavender reported that 41% of restraint episodes also involved use of medication and
4% resulted in seclusion. Other studies report that restraint episodes involved the
seclusion of patients in 10% (Leggett and Silvester 2003) and 13% (Parkes, 1996) of
cases. Riley et al (2006) found that patients restrained in the prone position were
more likely to be subject to high intensity observation after restraint than those in the
supine position. The authors acknowledge that this finding is difficult to explain, but
suggest that staff would find it more difficult to judge patients reactions to verbal
interventions whilst in the face down position. Although not directly comparable to
these studies, 12% of nurses surveyed from regional secure and psychiatric intensive
care units reported that the restraint incident in which they were last involved
required additional measures such as seclusion and medication were also used (Lee et
al 2003).
Characteristics of restrained patients
The characteristics of patients subject to restraint were generally poorly reported. It is
particularly difficult to interpret studies which report the characteristics of restrained

patients only, since it is not known how these patients characteristics differ from
those who were not restrained.
Age
Three studies of restrained patients report an average age of 31 (Leggett and Silvester
2003), 33 (Smith and Humphreys 1997) and 37 (Lancaster et al 2008). Older patients
are less likely to be restrained in a horizontal position (Whittington et al 2006).
Gender
Two studies reporting the gender of restrained patients found a greater proportion to
be male: 94% male (Duff et al 1996); 55% male (Lancaster et al 2008). Males also
comprised the majority (65%) of restraint requests (Ryan and Bowers 2006). In one
study, 77% of restrained patients were male, but comparatively more females (43%)
than males (31%) admitted during the study period were subject to restraint, although
this difference did not achieve statistical significance (Leggett and Silvester 2003).
On balance, therefore, more males than females may be subject to restraint, although
the quality of the evidence is weak.

Ethnicity
There is no consistent evidence that particular ethnic groups are more likely to be
restrained. Three studies found no association between ethnicity and the use of
restraint (Gudjonsson et al 2000, 2004; Duff et al 1996). The latter study reported
that 64% of restraint episodes involved Afro-Caribbean patients, but the same
proportion of patients from this ethnic group comprised the study sample of serious
violent incidents. Another study reported that 22% of restraint episodes involved
patients from an ethnic minority group and that this proportion was similar for men
and women (Lancaster et al 2008).
The Count Me In census reports provide conflicting evidence of ethnic differences.
In 2005, control and restraint among Black Caribbean men was 29% higher than the
average rate for all inpatients. In 2006, inpatients from the White/Black Caribbean
Mixed group were more likely than average to experience restraint, but there were no
ethnic differences among either men or women reported in 2007. In 2008, patients in
the Other White and White/Black Caribbean Mixed groups had a higher than average
rate of restraint (29% and 34% respectively).
Legal
Studies of patients in secure units, where most if not all patients are formally detained,
are excluded from this section. Among studies which reported the legal status of
patients there was a consistent finding that restrained patients were usually formally
detained. The proportion ranged from 82% (Duff et al 1996; Lancaster et al 2008) to
94% (Smith and Humphreys 1997). Ryan and Bowers (2006) reported a lower rate
(65%) for restraint requests, but did not provide data for patients who were actually
restrained. There is also evidence that patients are more likely to be restrained if on a
civil rather than criminal section (Gudjonsson et al 2000, 2004).
Diagnosis
Assessment of the relationship between restraint and diagnosis is made difficult by the
use of differing terminology and diagnostic systems between studies and variations in

the types of psychiatric services and their populations. Two studies report
schizophrenia to be the most common diagnosis for restrained patients (57%, Duff et
al. 1996; 60% Smith and Humphreys 1997). Lancaster et al.s (2008) study of 680
restraint episodes provides a more diverse range of diagnoses: schizophrenia (33%),
mania and excited psychosis (20%), paranoia and acute psychotic reaction (17%),
acute reaction and personality disorder (11%) and substance related or other (18%).
Where the relationship between diagnosis and restraint had been analysed statistically
no significant association was found (Gudjonsson et al 2004; Tobin et al 1991).
Length of stay
Only one study reported the length of stay for restrained patients (Smith and
Humphreys 1997). Half of restraint episodes involved patients who had been on the
wards for at least 3 weeks but half of the remaining episodes were for patients who
had been in hospital for less than 24 hours.
Other
Information on restrained patients marital status, employment, living arrangements,
educational qualifications and forensic history was not reported by any study.
Staff and patient experiences of manual restraint
This section deals specifically with the direct experience of restraint from the
perspective of staff (mostly nurses) and patients. The implications for staff training
are explored in the next section.
Staff
The most comprehensive study involved a postal questionnaire survey of 269 nurses
in regional secure and psychiatric intensive care units in England and Wales
concerning their last experience of using control and restraint (Lee et al 2003). Most
nurses (96%) reported positive outcomes of the restraint, but some negative aspects of
restraint (and of colleagues' attitudes) were identified and alternatives suggested. A
quarter or respondents expressed concerns about the impact on patients (e.g.
relationships with nursing staff, while some found the experience of restraint
demeaning and stressful. Organisational factors included the poor management of
restraint procedures, lack of monitoring and under-staffing. There were also doubts
about some of the techniques used (e.g. joint locks which induce pain to gain
compliance) and the impact restraint has on other patients. Most (70%) staff reported
de-briefing after the incident.
These findings are generally consistent with qualitative studies which suggest that
nurses view restraint as a necessary part of their job, but one they would like to
minimise (Bigwood and Crowe 2008; Bonner et al. 2002). Anxiety about getting hurt
and distress in implementing restraint were common themes (Bigwood and Crowe
2008; Bonner et al 2002; Sequira and Halstead 2004) reflecting a sense of conflict
with the therapeutic nursing role. One study found that nurses were reluctant or
unable to express their feelings following an episode of restraint, but that these
emotional responses diminished with greater experience of implementing restraint
(Sequira and Halstead 2004). Psychiatric nurses may be more willing to intervene in
aggressive incidents than their general nursing counterparts (Duxbury, 1999). The
positive benefits of debriefing after restraint episodes have also been acknowledged

(Bonner et al 2002). The use of seclusion after restraint has been linked to staff
perceiving patients to have control over the cause of the incident (Leggett and
Silvester 2003). The same study found that staff felt less control over incidents with
frequently restrained patients.
Patients
Less research is available on patients experiences of restraint. Two small qualitative
studies report a total of 20 patients experiences views of restraint (Bonner et al 2002;
Sequeira and Halstead 2002). Both found predominantly negative experiences
including feelings of anger, fear and panic. Patients said they felt ignored prior to the
incident and that their behaviour had not warranted the use of restraint. There was
also a consensus that restraint risked reawakening memories of previous distressing or
abusive events.
Staff training in restraint techniques
A survey of 33 violence management policies from Trusts across England and Wales
found that almost all specified physical restraint as a means of managing violence and
more than half that restraint should used as a last resort (Wright et al 2000). There
was more ambiguity about the role of untrained staff and the level of force
permissible. Less than half the policies set out when restraint would be justified,
unacceptable methods of restraint, checking for weapons before restraint and the
dignity of the patient during restraint.
Surveys of psychiatric nurses training suggest gaps in training provision, particularly
for techniques that might be used to avoid recourse to restraint, and that skills for
managing conflict are not updated regularly enough. Wright et al (2005) reported that
while most (77%) had received training in restraint techniques and some refresher
training during their current post, many had to wait several months before receiving it.
On the whole, nurses were not confident of applying restraint techniques or resolving
incidents without restraint. Lee et al.s (2001) survey of PICU and secure unit staff
also found some who had been in post several for months before receiving training.
Assuming that the dozen most frequently taught C&R techniques constitute a core
curriculum, less than two fifths of nurses had been trained in all core techniques. A
lack of training in safety and ethical issues was also highlighted. Another survey of
PICU staff also found that although the majority of staff identified management of
aggression and violence as their primary training need, it was felt that this should
include areas such as de-escalation, debriefing and seclusion as well as restraint
(Clinton et al., 2001).
There is some qualitative evidence that despite general satisfaction with their training,
staff are not always adequately prepared for de-escalating situations and dealing with
the most common forms of assaults such as punches and kicks (Southcott et al 2002).
Staff also reported having problems in taking patients to the floor and initial
establishment of holds. This is consistent with a study of Australian wards in which
some staff found restraint techniques complex and the training difficult to put into
practice, although aggression management training was the most commonly reported
factor giving clinicians confidence in dealing with patient aggression (Martin and
Daffern 2006).

The effectiveness of restraint training had been assessed by a number of studies which
measured staff skills and or confidence before and after participation in a training
programme. Paterson et al (1992) evaluated a 10-day in-service education course in
the management of violence and found significant improvements in knowledge,
stress, role ambiguity, and de-escalation and control and restraint skills, but not job
satisfaction. A survey of staff from three Australian psychiatric intensive care wards
measured staff confidence in dealing with aggressive patients (McGowan et al 1999).
At one, the survey was repeated six months after staff completed a safe physical
restraint training module. Staff at the other two wards had received regular training.
The survey showed that staff confidence at these two wards was higher than pretraining levels at the third. Following the training module, staff at the third hospital
showed a significant increase in confidence.
One study found that C&R training increased the modal number of staff required to
undertake restraints from two to three (a three staff team was emphasised in the
training), but perhaps as a consequence there was also and increase in staff injuries
during restraint (Parkes, 1996). Increasing the numbers of staff trained in C&R, to the
point when a team of three is available at all times, has been associated with a
reduction in the number and severity of violent incidents, although where incidents
did occur there was a tendency for patients to attack each other rather than nurses
(Mortimer, 1995).
Evidence for and against the working model
None of the reviewed studies provides evidence to support the working model. The
potential role of organisational support is a logical conclusion to some studies (e.g.
post-incident reviews for staff and patients) but is not explicitly explored. Although
the need for timely and comprehensive staff training in violence and aggression
management is often stated, the impact of training on the use of restraint and other
forms of containment has not been assessed. Instead, studies rely on measuring the
confidence and skills of staff following training. It is to be expected that staff feel
more confident after recent training, but it is much less clear how long this confidence
or improved competence lasts. In the absence of supporting data, it cannot be
assumed that improved training would reduce the incidence of restraint. It is equally
plausible that better trained staff would simply be more confident and able to restrain
patients safely. Increased familiarity with restraint techniques can lead to
intransigence or a hardening of attitudes towards the practice among some staff
(Sequeira and Halstead, 2004). More experienced staff may be more ready to
intervene or more willing to touch the patient first and less inclined to talk an agitated
patient down (Harris and Rice, 1986).
The subjective experience of staff is rarely reported in detail, but qualitative research
illustrates how difficult emotional regulation can be for some staff. One study in
particular found that some were unable to express their feelings following a restraint
episode and that this may contribute to less appropriate coping strategies such as
laughter after the restraint event, which patients find offensive (Sequeira and
Halstead, 2004). The study also reported staff anger when the same patients were
frequently restrained. Staff perceived these patients to be deliberately bringing about
a restraint episode. Similarly, Bigwood and Crowe (2008) report that nurses
participation in restraint made their jobs more difficult and reduced job satisfaction,
but they felt less conflict with their therapeutic role if satisfied that other options had

been properly explored before the use of restraint. These studies do not explain how
greater emotional regulation might translate into reduced use of restraint or other
containment methods to manage conflict.

Points the model has missed

Whether or not restraint is used may depend upon how staff use certain cues (e.g.
agitation and threat of violence) to decide how best to manage the patient
(Gudjonsson et al 2004). However, it is not clear which patient behaviours are
associated with particular staff responses and how they are interpreted by staff. For
example, interactions preceding the incident such as refusing a request may also be
important in determining what emergency actions will be taken. Staff may use
seclusion punitively as a means of regaining a feeling of control after a restraint
episode, but use help giving strategies when patients are perceived to have less control
over their circumstances (Leggett and Silvester, 2003). The decision to restrain a
patient on the floor seems to be associated with staff perceptions of an imminent
threat, perhaps based upon fear or a need to regain control (Whittington et al 2006).
There is also evidence that the choice of restraint technique may be influenced by
situational factors. An association between restraint in the prone position and a
warning of imminent violence may reflect staff having the opportunity to approach
the patient from the front and to attempt to de-escalate the situation face-to-face, prior
to the use of restraint which would be initiated from the front in the interests of patient
safety (Riley et al., 2006).
Patient variables were generally not associated with restraint, but gender appears to
have some bearing on staff-patient interactions concerning restraint. Leggett and
Silvester (2003) found use of medication following restraint to be associated with
staff attributing behaviour among male patients as uncontrollable. Female patients
were more likely to be secluded and less likely to receive medication, but staff had
more difficulty explaining the reasons for restraint episodes involving female patients.
Harris and Rice (1986) found that restraint episodes resulting in staff absence due to
injury were more likely to involve male staff and patients. The authors suggest that
male staff were more likely to initiate restraint or to be sought when restraint was
being planned, and female were patients less likely to need restraint. Although this
study provided no evidence to support this, others have reported male nurses to be
more willing to contribute to the restraint of an aggressive patient (Martin and
Daffern, 2006) and females to feel that restraint conflicts with their role as a nurse
(Sequeira and Halstead, 2004). One qualitative study found female patients who were
restrained by female staff expressed feelings of comfort or safety associated with
restraint, to the extent that they deliberately behaved ways which might provoke the
use of restraint (Sequeira and Halstead, 2002). These findings require cautious
interpretation, since the larger quantitative studies in this review show no association
between gender and restraint. This may reflect obvious differences in research
questions (prevalence vs staff-patient interactions), but also that when a range of
possible factors are considered together gender is of much less importance.
Discussion

Summary
Given the prevalence of manual restraint use across inpatient psychiatric services the
lack of data on this practice is striking. Even simple information about the frequency,
duration and reasons for restraint is hard to find. On the whole, it is difficult to draw
many conclusions about the use of restraint because the small number of studies
included cover a wide range of services treating diverse populations. The data
suggests that on an average 20 bed ward there might be between two and five restraint
episodes per month, with forensic services at the higher end of this range. Restraint is
not confined to the management of violent incidents, but is used to in response to a
range of patient behaviours. Whether staff are the target of assault and the legal status
of patients are also factors associated with restraint use. Restraint frequently precedes
sedation of patients (in around 40% of restraint episodes), but seclusion is used less
often. Staff value training in C&R techniques, not least because of the risk of injury
associated with violent incidents as well as the use of restraint. However, training is
not always timely and does not adequately cover safety and ethical issues. The
impact of training on levels of restraint use has not been evaluated. The extent to
which staff attitudes influence the decision to use restraint is uncertain. There was
some consistency across the qualitative studies that contextual factors play a role in
the management of incidents. Staff may empathise with patients who are frustrated
with the ward environment and consequently use less coercive interventions because
they feel able to communicate with the patient. In other circumstances, staff may find
it more difficult to negotiate with patients and resort to the use of restraint and/or
other containment methods.
Lessons for future research
There is an assumption in much of the literature, sometimes implicit in the research
design, that restraint is simply used to manage violent incidents. As this review has
shown, some of the more robust studies point to a range of behaviours and factors
associated with the use of restraint such as attempts to abscond, agitation, refusal to
comply with instructions, the target of assault and legal status. The range of
antecedents to restraint lead Ryan and Bowers (2006) to conclude that manual
restraint is not associated with violence, but the enforcement of detention and
treatment of patients.
A large proportion of the studies in the review used officially recorded data on
untoward incidents and the use of restraint. However, one study showed that the
recording of basic information about incidents was often omitted or inadequate
(Dowson et al 1999). This can have a detrimental effect on the management of
violence as failure to record incidents thoroughly means that information needed to
review processes and resources and assess high risk situations will be missed. The
potential for missing data and under-reporting means that results based upon officially
recorded data should be treated with caution. Other influences such as the concerns of
managers and changes in national and local policy may also undermine the reliability
and validity of the data. The diverse factors found to be associated with restraint
identified in the review suggest that future studies of official data should use
multivariate analysis wherever possible in order to avoid potentially misleading
conclusions based upon simple counts of events or antecedents. Bespoke data
collection instruments which capture more detailed information about the
management of incidents than currently available through official means would also

be welcomed. For example, it would be helpful to have more data on the timing of
events leading up to a restraint episode. Of course, these come with their own
problems and complications, especially the additional burden often placed on staff to
assist data collection.
Relatively few studies were specifically focused on restraint; most concerned topics
such as violent incidents or staff training. Again, this probably reflects the
availability of official data and all the limitations this implies. Yet, as this review has
highlighted, studies based upon restraint following violent incidents are not a reliable
indicator of the prevalence of the practice. A better sense of how restraint is used in
response to different types of incident and in different service settings is required. A
more thorough analysis of repeatedly restrained patients should also be a priority. It
is not at all clear how these individuals skew the available prevalence data or how the
circumstances of restraint episodes differ from other patients.
Now that prevention and management of violence and aggression training is
mandatory, there is a need for it to be properly evaluated. This necessarily means
measuring the impact of training on violent incidents and nursing responses to it.
Staff confidence and skills are important, but the literature does not offer much
information about how these have been put into practice. Whilst there is some
tentative evidence of a relationship between C&R training and reduced violence
(Mortimer 1995), more research is required to conclusively demonstrate such a link,
and it cannot be assumed that reduced violence would be mirrored by similar
reductions in episodes of restraint.

References
Bigwood, S. & Crowe, M. (2008). It's part of the job, but it spoils the job': a
phenomenological study of physical restraint. International Journal of Mental Health
Nursing, 17, 215-222.

Blofeld, J., Sallah, D., Sashidharan, S. P., Stone, R., Struthers, J. (2003). Independent
inquiry into the death of David Bennett. Norfolk, Suffolk and Cambridgeshire
Strategic Health Authority, Cambridge.

Bonner, G., Lowe,T., Rawcliffe, D., Wellman, N. (2002). Trauma for all: a pilot
study of the subjective experience of physical restraint for mental health inpatients
and staff in the UK. Journal of Psychiatric and Mental Health Nursing, 9, 465-473.

Bowers, L., Simpson, A., Alexander, J. (2003). Patient-staff conflict: results of a

survey of acute psychiatric wards. Social Psychiatry and Psychiatric Epidemiology,
38, 402-8.

Clinton, C., Pereira, S., Mullins, B.(2001). Training needs of psychiatric intensive
care staff. Nursing Standard, 15, 33-36.

Brooker, C. and Brabban, A. (2006). Effective training in psychosocial interventions

for work with people with serious mental health problems. The Mental Health
Review, 11, 7-14.

Cheung, P., Schweitzer, I., Tuckwell, v., Crowley, K.C. (1997). A prospective study
of assaults on staff by psychiatric inpatients. Medicine, Science and the Law, 37, 4652.
Department of Health (2008). Code of practice to the Mental Health Act 1983.
London: The Stationary Office.

Dowson, J. H., Butler, J., Williams, O. (1999). Management of psychiatric in-patient

violence in the Anglia region. Implications for record-keeping, staff training and
victim support. Psychiatric Bulletin., 23, 486-489.

Duff, L., Gray, R., Brostor, F. (1996). The use of control and restraint techniques in
acute psychaitric units. Psychiatric Care, 3, 230-234.

Duxbury, J. (1999). An exploratory account of registered nurses' experience of

patient aggression in both mental health and general nursing settings. Journal of
Psychiatric and Mental Health Nursing, 6, 107-114.

Gudjonsson, G., Rabe-Hesketh, S., Wilson, C. (2000). Violent incidents on a medium

secure unit: the target of assault and the management of incidents. Journal of
Forensic Psychiatry, 11, 05-118.

Gudjonsson,G., Rabe-Hesketh, S., Szmukler, G. (2004). Management of psychiatric

inpatient violence: patient ethnicity and use of medication, restraint and seclusion.
British Journal of Psychiatry, 184, 258-262.

Harris, G.T. & Rice, M.E. (1986). Staff injuries sustained during altercations with
psychiatric patients. Journal of Interpersonal Violence, 1, 193-211.

Healthcare Commission (2005). Count Me In. Results of a National Census of

Inpatients in Mental Health Hospitals and Facilities in England and Wales.
Healthcare Commission, London.

Healthcare Commission (2006). Count me in 2006. Results of the 2006 national

census of inpatients in mental health and learning disability services in England and
Wales. Healthcare Commission, London.

Healthcare Commission (2007). Count me in 2007. Results of the 2007 national

census of inpatients in mental health and learning disability services in England and
Wales. Healthcare Commission, London.

Healthcare Commission (2008). Count me in 2008. Results of the 2008 national

census of inpatients in mental health and learning disability services in England and
Wales. Healthcare Commission, London.

Kennedy, J., Harrison, J., Hillis, T., Bluglass, R. (1995). Analysis of violent incidents
in a regional secure unit. Medicine, Science and the Law, 35, 255-260.

Lancaster, G.A., Whittington, R., Lane, S., Riley, D., Meehan, C. (2008). Does the
position of restraint of disturbed psychiatric patients have any association with staff
and patient injuries? Journal of Psychiatric and Mental Health Nursing, 15, 306-312.

Larkin, E., Silvester, M., Jones, S. (1988). A preliminary study of violent incidents in
a speical hospital (Rampton). British Journal of Psychiatry, 153, 226-231.

Lee, S., Wright, S., Sayer, J., Parr, A., Gray, R., Gournay, K. (2001). Physical
restraint training in English and Welsh psychiatric intensive care and regional secure
units. Journal of mental Health, 10, 151-162.

Lee, S., Gray, R., Gournay, K., Wright, S., Parr, A-M., Sayer, J. (2003) Views of
nursing staff on the use of physical restraint. Journal of Psychiatric and Mental
Health Nursing, 10, 425-430.

Leggett, J. & Silvester, J. (2003). Care staff attributions for violent incidents
involving male and female patients: a field study. British Journal of Clinical
Psychology, 42, 393-406.

Martin, T. & Daffern, M. (2006). Clinician perceptions of personal safety and

confidence to manage inpatient aggression in a forensic psychiatric setting. Journal
of Psychiatric and Mental Health Nursing, 13, 90-99.

McGowan. S., Wynaden, D., Harding, N., Yassine, A., Parker, J. (1999). Staff
confidence in dealing with aggressive patients: a benchmarking exercise. Australian
and New Zealand Journal of Mental Health Nursing, 8, 104-108.

Morrison, A. & Sadler, D. (2001). Death of a psychiatric patient during physical

restraint. Excited delirium a case report. Medicine, Science & the Law, 41, 46-50.

Mortimer, A. (1995). Reducing violence on a secure ward. Psychiatric Bulletin, 19,

605-608.

National Institute for Mental Health in England (2004). Mental Health Policy
Implementation Guide. Developing Positive Practice to Support the Safe and

Therapeutic Management of Aggression and Violence in Mental Health In-patient

Settings. Department of Health, Leeds.

Needham, I., Abderbalden, C., Halfens, R., Fischer, J., Dassen, T. (2005). Nonsomatic effects of patient aggression on nurses: a systematic review. Journal of
Advanced Nursing, 49, 283-296.

National Institute for Clinical Excellence (2005). Violence: the short-term

management of disturbed/violent Behaviour in psychiatric in-patient settings and
emergency departments. National Institute for Clinical Excellence, London

Paterson, B., Leadbetter, D., McComish, A. (1998). Restraint and sudden death from
asphyxia. Nursing Times, 94, 62-64.

Parkes, J. (1996). Control and restraint training: a study of its effectiveness in a

medium secure psychiatric unit. Journal of Forensic Psychiatry, 7, 525-534.

Parkes, J. (2003). The nature and management of aggressive incidents in a medium

secure unit. Medicine, Science and the Law, 43, 69-74.

Paterson, B., Turnbull, J., Aitken, I. (1992). An evaluation of a training course in the
short-term management of violence. Nurse Education Today, 12, 368-375.

Paterson, B., Bradley, P., Stark, C., Saddler, D., Leadbetter, D., Allen, D. (2003).
Deaths associated with restraint use in health and social care in the UK. The results of
a preliminary survey. Journal of Psychiatric and Mental Health Nursing, 10, 3-15.

Powell, G., Caan, W., Crowe, M. (1994). What events precede violent incidents in
psychiatric hospitals? British Journal of Psychiatry, 165, 107-112.

Riley, D., Meehan, C., Whittington, R., Lancaster, G.A., Lane, S. (2006). Patient
restraint positions in a psychiatric inpatient service. Nursing Times, 3, 42-45.

Royal College of Psychiatrists (2007). Healthcare Commission National Audit of

Violence 2006-7: Final Report. Royal College of Psychiatrists, London

Ryan, C. & Bowers, L. (2006). An analysis of nurses' post-incident manual restraint

reports. Journal of Psychiatric and Mental Health Nursing, 13, 527-532.

Sailas E, Fenton M. (2000). Seclusion and restraint for people with serious mental
illnesses. Cochrane Database of Systematic Reviews, Issue 1. Art. No.: CD001163.
DOI: 10.1002/14651858.CD001163.

Sequeira, H. & Halstead, S. (2002). Control and restraint in the UK: Service user
perspectives. British Journal of Forensic Practice, 4, 9-18.

Sequeira, H. & Halstead, S. (2004). The psychological effects on nursing staff of

administering physical restraint in a secure psychiatric hospital: 'when I go home, it's
then that I think about it'. British Journal of Forensic Practice, 6, 3-15.

Shepherd, M. & Lavender, T. (1999). Putting aggression into context: an

investigation into contextual factors influencing the rate of aggressive incidents in a
psychiatric hospital. Journal of Mental Health, 8, 159-170.

Smith, A.D. & Humphreys, M. (1997). Physical restraint of patients in a psychiatric

hospital. Medicine, Science and the Law, 37, 145-149.

Southcott, J. Howard, A., Collins, E. (2002). Control and restraint training in acute
mental health care. Nursing Standard, 16, 33-36.

Southcott, J. & Howard, A. (2007). Effectiveness and safety of restraint and

breakaway techniques in a psychiatric intensive care unit. Nursing Standard, 21, 3541.

Tobin, M., Lim, L., Falkowshi, W. (1991). How do we manage violent behaviour?
British Journal of Clinical and Social Psychiatry, 8, 19-23.

Torpy, D. & Hall, M. (1993). Violent incidents in a secure unit. Journal of Forensic
Psychiatry, 4, 517-544.

Whittington, R., Lancaster, G., Meehan, C., Lane, S., Riley, D. (2006). Physical
restraint of patients in acute mental health care settings: patient, staff and
environmental factors associated with the use of a horizontal restraint position.
Journal of Forensic Psychiatry & Psychology, 17, 253-265.

Wright, S., Sayer, J., Parr, A.-M., Gray, R., Southern, D., Gournay, K. (2005).
Breakaway and physical restraint techniques in acute psychiatric nursing. Journal of
Forensic Psychiatry and Psychology, 16, 380-398.