PRiSM, Institute of Psychiatry London
Correspondence: Jonathan Bindman, Section of Community Psychiatry (PRiSM), Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE5 8AF
Declaration of interest This study was funded by a grant from the Department of Health. The views expressed are those of the authors and not necessarily those of the Department of Health.
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Aims To identify the factors associated with registration, and obtain clinicians' views on its effectiveness.
Method At a random sample of 14 trusts data were collected from case notes, keyworkers and responsible medical officers.
Results A sample of 133 registered patients were more disabled and had more extensive histories of violence and self-harm than 126 comparison patients on the upper tier of the Care Programme Approach (CPA). Those registered were a heterogeneous group. For some there was little evidence of risk. In most cases clinicians did not believe registration had improved care.
Conclusions The Supervision Register policy has not resulted in the identification of a well-defined group. Its effectiveness is limited by the lack of operationalised measures of risk.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We aimed to : (a) establish the characteristics of individuals included on Supervision Registers ; (b) compare these with patients requiring a high level of care but not regarded as at high risk ; and (c) assess the views of responsible medical officers (RMOs) and keyworkers about the impact of inclusion on the Register.
![]() |
METHOD |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Case notes of all cases and comparison patients were examined by a researcher (A.B. or J.B.), and demographic and service use data were recorded. Previous incidents of violence and self-harm recorded in case notes (lifetime most severe, and most severe in the last six months) were extracted and recorded on a form designed for the purpose. Notes were read in detail, and the researcher could not be blinded to the patient's Supervision Register status. Incidents were rated for severity on a scale of 0-4 using the anchor points on the overactive, aggressive, disrupted or agitated behaviour and non-accidental self-injury scales included in the Health of the Nation Outcome Scales (HoNOS ; Wing et al, 1998). Each form was rated by two researchers (A.B. and J.B.), and a consensus rating made in cases of initial disagreement, which was by no more than a single point in any case. Incidents of life threatening violence (either described as such in case notes, or assaults involving stabbing or strangulation), were rated in addition. References in case notes to self-neglect were also recorded and rated as : neglect resulting in threat to life, and neglect resulting in failure to obtain adequate nourishment. The RMO of each Supervision Register case was asked to complete a questionnaire concerning the reasons for placing the patient on the Supervision Register, and the RMO's view of the effect of inclusion on the Supervision Register on the patient's care. The keyworkers of all Register cases and all CPA comparison patients were interviewed to provide current ratings (the month prior to rating, or the most recent contact if longer than a month) using the HoNOS, the Camberwell Assessment of Need (CAN ; Phelan et al, 1995), and the Global Assessment of Functioning (GAF) symptom and disability scales (Endicott et al, 1976). They were also asked at interview to complete a questionnaire concerning their views of the effect of inclusion on the Supervision Register on the management of each case.
Data analysis
Data were entered into a database and analysed using STATA for Windows
(STATA Corporation, 1997).
Descriptive statistics were used for the survey results, and Supervision
Register cases and CPA comparison patients were compared using
X2 and t-tests for categorical and continuous
variables as appropriate, and a non-parametric test for trend (nptrend) for
the ordered categorical variables resulting from the HoNOS ratings of violence
and self-harm.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
As shown elsewhere (Bindman et al, 1999), variations in local need do not explain the variation in the numbers, which instead reflect the application of inconsistent criteria for inclusion.
Characteristics of patients registered
Of 140 Supervision Register cases sampled as described above, data were
obtained on 133 (95%), and of the 133 comparison patients, matched for RMO,
data was obtained on 126 (95%). The demographic and service use
characteristics of patients and comparison patients are shown in
Table 1.
|
Keyworkers' ratings of Supervision Register cases and CPA comparison patients on the HoNOS, GAF and CAN scales are shown in Table 2. The GAF scores show that as a group Supervision Register cases have high levels of symptoms and disability, significantly greater than comparison patients. The total HoNOS score confirms this finding. Examining HoNOS item scores, the Supervision Register cases are rated as having higher levels of overactive/aggressive behaviour, hallucinations and delusions, and physical illness or disability, when compared with comparison patients (P <0.01, X2 test for trend). The overall number of needs, as measured by the CAN, is significantly higher among Supervision Register cases, as is the number of met needs. This would support the view that patients on Registers are correctly identified as being those in greatest need, and that services are responding appropriately to their needs. However, the absolute difference in total and in met needs between the Supervision Register and comparison groups is small, and the Supervision Register cases also have slightly higher unmet needs, though this difference is significant only at the 10% level.
|
Incidents of aggression, self-harm, and self-neglect
Structured assessments of risk were largely absent from case notes of
Register cases, and in only 50 cases (37%) was any justification of inclusion
on the Supervision Register (other than a simple statement of entry and
category of risk) recorded. Fig.
2 shows that the risk of violence, alone or in combination with
other categories, was the most common reason for placing patients on the
Supervision Register, and risk of selfharm was the least common. The
relatively high use of the self-neglect category, despite its lower profile as
an area for concern by psychiatric services, suggests that it may be regarded
by psychiatrists as more predictable, or easier to prevent, than violence or
suicide.
|
History of aggression in patients registered as a risk to
others
Of the 133 Supervision Register cases, 81 (61%) were registered as being at
risk of harming others (32 of whom were also in other risk categories). In 14
of these cases (17%) there was no history of any aggression recorded in case
notes (equivalent to HoNOS score 0), and in a further 14 (17%) only verbal
aggression was recorded (HoNOS score 2). Eight patients (10%) had behaved in a
threatening way (HoNOS score 3), and 45 (56%) had committed an actual physical
assault equivalent to a HoNOS rating of 4. Few of the recorded incidents were
recent, 51 cases (63%) having no recorded aggression within the last 6 months,
and only nine (11%) having committed an actual assault (HoNOS score 4).
The Supervision Register cases had significantly higher levels of recorded aggression (z=-7.6, P < 0.001) than the CPA comparisons (of whom 15 cases, 12%, had committed an actual assault, equivalent to a HoNOS rating of 4). In addition, of the Register cases who were recorded as committing assaults, nine had committed life-threatening assaults, of whom one had committed homicide many years previously. One of the comparison group had committed a life-threatening assault and one had committed homicide many years previously and was subject to restriction under Section 49 of the Mental Health Act 1983.
Patients categorised as at risk of self-harm
Forty-three patients were registered in the category of risk of self-harm
(of whom 27 were also in other categories). Rating incidents recorded in case
notes using the equivalent of the HoNOS self-harm rating, 10 cases (23%) had
no record of any lifetime suicidal thoughts or behaviour, and a further five
(11%) only minor risk (equivalent to a HoNOS rating of 1 or 2). Seven cases
(16%) were recorded as having been regarded as at moderate risk (HoNOS score
of 3), and 21 (49%) had incidents of actual self-harm (HoNOS score 4) recorded
in case notes. Few incidents were recent, only four cases (9%) having made a
serious attempt at self-harm (HoNOS score of 4) in the previous six
months.
Though Supervision Register cases had significantly higher levels of recorded self harm (z=-4.7, P < 0.001), 23 (18%) of the comparison group had made recorded suicide attempts rated as a HoNOS score of 4. As for aggression, though the Register cases as a group are at higher risk than the comparison group, many patients with no recorded history of actual self-harm are included in this supposedly very high risk group, and risk assessment in this area is again inconsistent.
Patients categorised as at risk of self-neglect
There is no standard method of assessing self-neglect, but we were able to
identify from case notes that of 45 patients registered as at risk of
self-neglect (18 of whom were in other categories as well), 16 (36%) had been
noted as at some time suffering weight loss as a result of failing to obtain
adequate food, compared with 20 (16%) in the comparison group, and that seven
(16%) cases on the Register had been recorded as neglecting themselves to a
life-threatening extent compared with 12 (10%) in the comparison group
(z=-3.8, P < 0.001).
Keyworker and RMO opinions of the effectiveness of the Supervision
Register
The keyworkers of 111 (83.5%) of the cases on the Supervision Register
responded to questions about their views of the effectiveness of the Register
in ensuring prioritisation, communication about risk, and preventing loss to
follow-up. After repeated mailings and telephone contacts, a total of 37 RMOs
answered similar questions in relation to 92 (69%) of the Register cases. All
responses were rated as positive or negative. The keyworkers believed that in
81 cases (73%) the patients had not been prioritised to receive more services
than they otherwise would have done, in 77 cases (69%) the risk of loss to
follow-up had not been reduced, and in 70 cases (63%) information sharing had
not improved as a consequence of inclusion on the Supervision Register. The
RMOs believed that in 68 cases (74%) placing the patient on the Supervision
Register had not had any benefits. In the 24 cases (26%) where benefits were
believed to have resulted, these included improved prioritisation, follow-up
and communication, but also consequences such as influencing discharge from
restriction or reassuring patients' relatives, which may not have been
associated with direct benefits for the patient.
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Criteria of risk
The study is limited by the lack of any gold standard measures of
individual risk, and in choosing to measure recorded incidents of risk
behaviours, we have used the only available proxy for actual risk.
The difficulty of measuring individual risk underlay the concern expressed at the introduction of the Supervision Registers, that the criteria for entry to the register were unclear (Caldicott, 1994 ; Harrison, 1994). Prior to the introduction of the Supervision Registers, some estimates of the numbers of patients likely to be registered used broad criteria of risk and large registers were predicted, from between 40 and 160 patients per 100 000 total population (Pugh et al, 1994 ; Laugharne, 1994) to 300 people per 100 000 total population (Caldicott, 1994). Though no formal target was suggested by the Department of Health, they suggested instead that narrow criteria should be adopted (Glover et al, 1994), and that these might result in the placing on the Supervision Register of 15 cases per 100 000 (Holloway, 1994). In practice, though even the largest registers have not approached the highest predictions, and the average is well below the lowest, there is evidence that both wide and narrow risk criteria are in use in different trusts. In the absence of sensitive and specific tests of risk, the adoption of either broad or narrow criteria is likely to be ineffective. Using broad criteria many lowrisk individuals will be included, wasting resources or causing the measure to be regarded as a mere paper exercise, and using narrow criteria a few token cases with extreme histories will be selected and the impact on the overall level of risk, to patients or to the community, will be small. However, such variable application of the policy is inevitable, given the lack of operationalised definitions of risks, and of sensitive, specific instruments for assessing them (Monahan, 1994 ; Gunnell & Frankel, 1994), particularly for non-forensic populations.
Views of effectiveness of the policy
In practice it appears that there is considerable doubt about the
effectiveness of the policy in improving care for individual patients, a
majority of keyworkers and RMOs regarding the policy as not having had the
positive impact intended.
Implications
The CPA policy has recently been reviewed, and trusts will be permitted to
discontinue Supervision Registers from April 2001 if a simplified two-tier
CPA, of which "risk assessment is an ongoing and essential part"
is functioning effectively (Department of
Health, 1999). This may make it possible for individuals to
receive the benefits of prioritisation without stigmatisation. However, while
removing Supervision Registers and integrating risk assessment further into
the CPA may make the lack of consensus about risk assessment suggested by this
study less apparent, it will remain a problem. The Mental Health Act 1983 is
currently under review, and new measures aimed at high-risk individuals in
community-based psychiatric populations are being considered
(Dobson, 1998). It seems
probable that most clinicians will identify only a small number of individuals
as at risk in response to policies emphasising risk assessment
(Bindman et al, 1999 ;
Pinfold et al, 1999),
and given the low specificity of risk prediction, the overall impact on risk
in the community is likely to be limited
(Shaw et al, 1999).
The possibility remains, however, that in some services large numbers of
low-risk individuals may be targeted, with implications for civil liberties
and for equity of service provision.
![]() |
CLINICAL IMPLICATIONS AND LIMITATIONS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
LIMITATIONS
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Caldicott, F. (1994) Supervision Registers : the College's response. Psychiatric Bulletin, 18, 385-386.
Department of Health (1995) Health of the Nation : Building Bridges. A Guide to Interagency Working for the Care and Protection of Severely Mentally III People. London : Department of Health.
(1999) Effective Care Co-Ordination in Mental Health Services. Modernising Mental Health Services. London : Department of Health.
Dobson, F. (1998) Frank Dobson outlines third way for mental health. http://www.coi.gov.uk/coi/depts/GDH
Endicott, J., Spitzer, R. L., Fleiss, J. L., et al (1976) The Global Assessment Scale. A procedure for measuring overall severity of psychiatric disturbance. Archives of General Psychiatry, 33, 766-771.[Abstract]
Glover, G., McCulloch, A. W. & Jenkins, R.
(1994) Supervision Registers for mentally ill people.
British Medical Journal,
309,
809-810.
Gunnell, D. & Frankel, S. (1994) Prevention
of suicide : aspirations and evidence. British Medical
Journal, 308,
1227-1233.
Harrison, K. (1994) Supervision in the community. New Low Journal, 144, 1017.
Holloway, F. (1994) Supervision Registers. Recent government policy and legislation. Psychiatric Bulletin, 18, 593-596.
House of Commons Health Committee (1993) Community Supervision Orders. Health Committee 5th Report. Vol. 1. London : HMSO.
Langharne, R. (1994) Most patients in Bow and
Poplar would be on the register. British Medical
Journal, 309,
1159.
Monahan, J. (1994) Towards a rejuvenation of risk assessment research. In Violence and Mental Disorder. Developments in Risk Assessment (eds J. Monahan & J. Steadman), pp. 1-17. Chicago, IL : University of Chicago Press.
NHS Executive (1994) Introduction of Supervision Registers for Mentally III People from 1 April 1994. (HSG(94)5.) Leeds : NHS Executive.
Phelan, M., Slade, M., Thornicroft, G., et al (1995) The Camberwell Assessment of Need : the validity and reliability of an instrument to assess the needs of people with severe mental illness. British Journal of Psychiatry, 167, 589-595.[Abstract]
Pinfold, V., Bindman, J., Friedli, K., et al (1999) Supervised Discharge Orders in England : compulsory care in the community. Psychiatric Bulletin, 23, 199-203.
Pugh, R., Gardner, J. & Allen, R. (1994)
Implications of Supervision Registers in Psychiatry. British
Medical Journal, 309,
611.
STATA Corporation (1997) STATA Statistical Software : Release 5.0. College Station, TX : STATA Corporation.
Shaw, J., Appleby, L., Amos, T., et al
(1999) Mental disorder and clinical care in people convicted
of homicide : national clinical survey. British Medical
Journal, 318,
1240-1244.
Wing, J. K., Beevor, A. S., Curtis, R. H., et al (1998) Health of the Nation Outcome Scales (HoNOS) : research and development. British Journal of Psychiatry, 172, 11-18.[Abstract]
Received for publication June 17, 1999. Revision received October 7, 1999. Accepted for publication October 12, 1999.