Pendulum management in secure services
The British Journal of Psychiatry 2004 184: 270-271
Access the most recent version at doi:10.1192/bjp.184.3.270-a
Reprints/ To obtain reprints or permission to reproduce material from this paper, please
permissions write to email@example.com
You can respond http://bjp.rcpsych.org/cgi/eletter-submit/184/3/270-a
to this article at
Email alerting Receive free email alerts when new articles cite this article - sign up in the box at
service the top right corner of the article or click here
Downloaded bjp.rcpsych.org on May 12, 2011
from Published by The Royal College of Psychiatrists
To subscribe to The British Journal of Psychiatry go to:
B R I T I S H J O U R N A L O F P S YC H I AT RY ( 2 0 0 4 ) , 1 8 4 , 2 7 0 ^ 2 7 4
Correspondence University of Auckland, New Zealand. E-mail:
EDITED BY STANLEY ZAMMIT A.Venneri Department of Psychiatry,University
Contents & Smaller trials for better evidence & Pendulum management in secure
services & Sertraline and exposure therapy in social phobia & Premature conclusions
about depression prevention programmes & Homicide data & Modest but growing
presence of Brazil in mental health and psychiatric research Pendulum management in secure
Smaller trials for better evidence nature and the modalities of efficacy using Tilt (2003) defends himself clearly against
cholinesterase inhibitors. In the same way, the criticisms of Drs Exworthy & Gunn
The interesting debate between Parker and studies of smaller groups of patients receiv- (2003). However, he does not emphasise
Anderson & Haddad (2003) suggests more ing treatment for depression may reveal the extent to which they have misrepre-
fundamental reasons to question prevailing correlations between clinical features and sented aspects of the Tilt Report (Tilt et
research paradigms and designs in respect treatment responses that are more likely al,
al, 2000). Specifically, Exworthy & Gunn
of the efficacy of and indications for psy- to guide the selection of therapy for individ- state, following their quote from the Report
chotropic medicines. That the clinical trial ual patients (Mayberg, 2003). concerning the relationship between secur-
industry reveals even marginal drug effects Large randomised controlled trials, by ity and therapy, ‘one should go further
may be seen as surprising given the virtual submerging variation in the interest of mar- because in high secure hospitals therapy in
absence of any basis for a taxonomy of ginal statistical significance, seem to offer its widest sense is an integral part of
mental disorders, other than the syndromal limited hope of significantly improving the security’. This virtually paraphrases the
classifications used in psychiatric practice. evidence that guides clinical practice. Stu- Report itself: ‘Security is the responsibility
There is little evidence that the major syn- dies of cognitive and pharmacological inter- of all personnel in a high security hospital
dromes align with any readily defined ventions might best be carried out with and . . . good security facilitates good
pathophysiological variance. Group hetero- smaller patient groups for whom there has therapy and vice versa’ (paragraph 8.2).
geneity in trial work, as the debaters remark, been detailed assessment of relevant patho- There also appears to be a marked
will therefore attenuate the evidence for sub- physiological and cognitive variance, as absence from this debate of both historical
stantial drug treatment effects, sometimes to well as the manifest clinical symptoms. and organisational perspectives. Rapoport
vanishing point. Meta-analysis of such data (1960) suggested, in considering the insti-
may not be much more revealing, com- Declaration of interest tutional dynamics of therapeutic institu-
pounding the influence of variable sampling tions, that ‘disturbances were partly a
M.F.S. and A.V. have received honoraria
in individual trials and publication bias. function of cycles of abdication of author-
and support for attending scientific meet-
These side-effects of the randomised ity, in the name of permissiveness, followed
ings, been members of advisory boards
controlled trial ethos are not greatly miti- by authoritarianism to restore order’. The
and received research grants from compa-
gated in the field of organic mental disor- consequences of the report on the Ash-
nies involved in the manufacture and
ders. At huge expense, multicentre trials worth Hospital inquiry (Blom-Cooper et
marketing of cholinesterase inhibitors.
of cholinesterase inhibitors in patients clas- al,
al, 1992) (Ashworth at that time being an
sified as probably having Alzheimer’s dis- abusive, authoritarian institution) were
Lanctot, K. L., Herrman, N.,Yau, K. K., et al (2003)
ease have shown only very modest (and to Efficacy and safety of cholinesterase inhibitors in clearly thought by Fallon et al (1999) to
many observers still unconvincing) effects Alzheimer’s disease: a meta-analysis. Journal of the relate to a breakdown of security (permis-
on cognitive and neuropsychiatric out- Association, 169,
Canadian Medical Association, 169, 557^564.
siveness), leading to the Tilt Report (which
comes (e.g. Lanctot et al, 2003). This is
ˆ MacKeith, I. G., Del Ser, T., Spano, P. F., et al (2000) has been perceived by many in secure
because these conditions, pace distin- Efficacy of rivastigmine in dementia with Lewy bodies: a services as authoritarian).
randomised, double-blind, placebo-controlled
guished efforts at nosological definition in Perhaps attempting to understand this
international study. Lancet, 356, 2031^2036.
life, are also heterogeneous. This variabil- cycle more, and how it may relate to the
ity, already evident from detailed clinical Mayberg, H. S. (2003) Modulating dysfunctional limbic-
cortical circuits in depression: towards development of
complex (and potentially contradictory)
and neuropsychological assessment, is brain-based algorithms for diagnosis and optimized tasks facing secure psychiatric services, might
further revealed by functional and structur- treatment. British Medical Bulletin, 65, 193^207.
Bulletin, 65, reduce the likelihood of yet more scandals,
al analysis of the brain. It is these data Parker, G./Anderson, I. M. & Haddad, P. (2003) inquiries and reports in the future. Scott
which might best inform sampling for Clinical trials of antidepressant medications are (1975) suggested that ‘detaining custodial
therapeutic trials. Studies on a smaller producing meaningless results (debate). British Journal of institutions have two aims, one therapeutic,
Psychiatry, 183, 102^104.
scale, therefore, targeting the more readily the other custodial. These can and should
defined Lewy body dementia (e.g. McKeith Venneri, A., Shanks, M. F., Staff, R. T., et al (2001) be complementary, but there is a tendency
et al, 2000) or more intensively charac-
al, Cerebral blood flow and cognitive responses to
for these functions to polarise out and
rivastigmine treatment in Alzheimer’s disease.
terised and monitored patients with NeuroReport, 13, 83^87.
NeuroReport, 13, eventually split like a living cell into two
Alzheimer’s disease (e.g. Venneri et al, al, separate institutions’. The debate between
2001) in both double-blind and open-label M. F. Shanks Department of Psychiatry, Faculty Exworthy & Gunn and Tilt illustrates the
designs, can convincingly demonstrate the of Medical and Health Sciences, Private Bag 92019, recurring nature of this phenomenon.
C O R R E S P ON D E N C E
Perhaps this debate needs to move on to a responders compared with 33.0% of the GmbH, Bristol-Myers-Squibb, Eli Lilly and
creative engagement with this process. patients treated with exposure plus pla- Company, GlaxoSmithKline, Janssen-Cilag,
cebo. I wonder why it was not mentioned Lundbeck, Meiji-Seiko Pharmaceuticals,
Blom-Cooper, L., Brown, M., Dolan, R., et al (1992) in the second paper whether the three active Novartis Pharmaceuticals Corp., Organon,
Report of a Committee of Inquiry into Complaints about
groups differed from placebo and from Sanofi-Synthelabo,
Pfizer Inc., Roche, Sanofi-Synthelabo,
Ashworth Hospital. London: HMSO.
each other on the primary efficacy Solvay, and Wyeth Pharmaceuticals.
Exworthy, T. & Gunn, J. (2003) Taking another tilt at
high secure hospitals. The Tilt Report and its
consequences for secure psychiatric services. British Instead, Haug et al report only relative Blomhoff, S., Haug, T. T., HellstrÖm, K., et al (2001)
Journal of Psychiatry, 182, 469^471.
, changes of mean scores without adjusting Randomised controlled general practice trial of
sertraline, exposure therapy and combined treatment in
Fallon, P., Bluglass, R., Edwards, B., et al (1999) for the large absolute differences at termi- Psychiatry,
generalised social phobia. British Journal of Psychiatry,
Report of the Committee of Inquiry into the Personality nation of the acute study (week 24). After 179,
Disorder Unit, Ashworth Special Hospital. London: 52 weeks, exposure patients only caught
Stationery Office. Haug, T. T., Blomhoff, S., HellstrÖm, K., et al (2003)
up to the already better scores of the sertra- Exposure therapy and sertraline in social phobia: 1-year
Rapoport, R. (1960) Community as Doctor. London: line groups. From both papers, I calculated follow-up of a randomised controlled trial. British Journal
Social Science Paperbacks. Psychiatry 182,
of Psychiatry, 182, 312^318.
the following total mean changes for weeks
Scott, P. D. (1975) Has Psychiatry Failed in the Treatment 0–52 by adding the mean changes for
of Offenders? (The Fifth Denis Carroll Memorial B. Bandelow Department of Psychiatry and
Lecture). London: Institute for the Study and Treatment
weeks 0 to 24 and the ones for weeks 24
Psychotherapy, The University of Gottingen, von-
of Delinquency. to 52 and found: 1.68 for placebo, 2.02
Siebold-Str. 5, D-37075 Gottingen,Germany.
Tilt, R. (2003) High-security hospitals (letter). British
for sertraline plus exposure, 1.92 for sertra-
Journal of Psychiatry, 182, 548.
, line, and 1.88 for exposure plus placebo on
Tilt, R., Perry, B., Martin, C., et al (2000) Report of the
the CGI–SP overall severity. For the SPS, I
Review of Security at the High Security Hospitals. London: found the following mean changes: 12.09 Author’s reply: The primary efficacy mea-
Department of Health. for placebo, 15.56 for sertraline plus expo- sures from our paper about treatment effect
sure, 14.12 for sertraline, and 15.91 for ex- al,
at week 24 (Blomhoff et al, 2001) are re-
D. Beales Mersey Care NHS Trust, Ashworth posure plus placebo. These scores may ported in the method section of the paper
Hospital, Parkbourn, Maghull, Merseyside L31 1BD, change a little bit after correction for parti- about the follow-up study (Haug et al, al,
and Bolton, Salford and Trafford Mental Health NHS cipants who withdrew from the trial. I 2003). In the pairwise comparisons, com-
doubt that any of these scores differs signif- bined sertraline and exposure and sertraline
icantly from each other or from placebo. By alone were significantly superior to pla-
no means is it true that ‘Exposure therapy cebo, while a non-significant trend towards
Sertraline and exposure therapy given alone is more effective in the long increased efficacy of exposure alone
in social phobia term than when given in combination with compared with placebo was reported.
I read with interest the article by Haug et al sertraline’. The opposite is the case: it takes The four study groups had a significant
(2003), but was puzzled by the conclusion 1 year for the exposure patients to reach the reduction in scores on all social phobia
they drew from their data. level of improvement that the sertraline and scales from baseline to follow-up. Further-
After a 24-week study comparing ser- the combination patients have already more, there was no significant difference
traline, sertraline plus exposure, exposure reached after half a year. Perhaps the in scores on primary efficacy measures be-
plus placebo, and placebo in patients with patients treated with exposure only showed tween the active treatment groups in any
social anxiety disorder (Blomhoff et al, al, further improvement during the ‘treatment- of the time-point analyses between week 0
2001), patients were followed up at week free’ follow-up period because one-fifth of and week 24. In the follow-up analyses
52. In the summary the authors conclude them now received treatment with selective we were therefore mainly interested in the
that ‘Exposure therapy alone yielded a serotonin reuptake inhibitors. Remarkably, changes after cessation of treatment. For
further improvement during follow-up, there was no deterioration in the sertraline the exposure group and the placebo group
whereas exposure therapy combined with groups on the primary efficacy measures, there was a further improvement in scores
sertraline and sertraline alone showed a despite the fact that only one-fifth of this on social phobia from week 24 to week
tendency towards deterioration after the group remained on medication. 52 and the changes on several of the sub-
completion of treatment’. This seems to be I have calculated a Bonferroni- scales were highly significant. On SF–36,
a misleading interpretation of their data. corrected critical P-value of 0.0073 when which demonstrates changes in a more
Haug and colleagues did not mention seven scales are used. Thus, all P-values global functioning, there was a significant
the primary efficacy measures of their study 50.05 and 50.01 given in the paper may improvement for the exposure alone and
in their paper. Reading the original paper be not significant. the placebo groups, while there was a
by Blomhoff et al, I find that the primary
al, I would suggest that the authors analyse significant deterioration in both the
efficacy measures were numbers of respon- their primary efficacy measures and sertraline-treated groups. Changes in scores
ders or partial responders on the Clinical reinterpret their data. on other social phobia scales for the
Global Impression – Social Phobia (CGI– sertraline-treated groups were non-
SP) and the Social Phobia Scale (SPS). In significant, but there was a tendency
the first study, treatment with sertraline Declaration of interest towards deterioration (Tables 1 and 2,
was superior to placebo, but exposure was B.B. is or has been a speakers’ bureau parti- pp. 314–315). We agree that the changes
not. For example, 45.5% of the patients cipant with Aventis, AstraZeneca Pharma- in sertraline-treated groups during the
treated with sertraline plus exposure were ceuticals, Bayer AG, Boehringer-Ingleheim follow-up period were marginal. However,
C O R R E S P ON D E N C E
contrasting these minimal changes with the of the Coping with Depression Course by Home Office statistical bulletins and the
significant improvement in the exposure- Lewinsohn et al (1984) were coded differ- House of Commons Library. For exam-
treated group, we find it appropriate to ently (e.g. ‘cognitive and competence’, ‘be- ple, Richards (1999) describes homicide
conclude that exposure therapy given alone havioural, cognitive, educational and trends between 1945 and 1997, demon-
seems to be more beneficial in the long social support’, ‘cognitive’, and ‘behaviour- strating the dramatic rise in rates of
term. Longer follow-up could have added al, cognitive, competence and educational’ offences initially recorded as homicide
valuable information to this issue. In all (pp. 386–391)). Finally, the coding category seen over that time from around 300 or
groups about 20% of the patients were ‘behavioural methods’ incorporates very 400 a year in the 1950s to more than
treated with sertraline during the follow- heterogeneous strategies. For example, 700 a year in the late 1990s. The recent
up period so this could not explain the behavioural strategies found to be helpful Home Office Statistical Bulletin (Simmons
differences in scores between the groups at in cognitive–behavioural therapy for de- & Dodd, 2003) shows a continuing rise
week 52. pression focus on increasing pleasant activ- in this trend with 1048 deaths initially
ities and social skills training (Lewinsohn et attributed to homicide in 2002/2003,
Declaration of interest al,
al, 1984), whereas the delivery of peer sup- although these figures are based on date
port telephone dyads by lay persons, as used of notification and thus can include
Funding was provided by Pfizer, Inc.
in the studies by Heller et al (1991), may be deaths that actually took place in earlier
T. T. Haug University of Bergen, Department of
regarded as a very specific behavioural years.
Psychiatry, Section Haukeland University Hospital, strategy which has so far not been recom- Dr Salib’s paper appears to use data on
N-5021 Bergen, Norway mended as a helpful intervention by the re- death registrations from the ONS where
search community. In Jane-Llopis et al’s
´ al’s there has been a conviction for murder or
meta-analysis, respective interventions from for manslaughter. However, the ONS as-
the studies by Heller et al (1991) had nega- signs a temporary ICD–9 code for cause
Premature conclusions about tive effect sizes and therefore may have of death for deaths where death was vio-
depression prevention programmes substantially accounted for the missing or lent, unnatural or suspicious or pending
In my opinion, the meta-analysis by Jane-
´ even negative effect of the ‘behavioural’ the outcome of inquests and legal proceed-
Llopis et al (2003) suffers from some meth- component of preventive measures. ings, which are of course often prolonged.
odological flaws that misguided the authors The ONS site itself states that it is difficult
to draw premature conclusions on predic- Bisson, J. I., Jenkins, P. L., Alexander, J., et al (1997) to present accurate statistics on number of
Randomised controlled trial of psychological debriefing
tors of prevention in depression prevention homicides using death registrations, which
for victims of acute burn trauma. British Journal of
programmes. Psychiatry, 171, 78^81.
, is what Dr Salib has seemingly attempted
First, many of the selected studies did to do.
Heller, K.,Thompson, M. G.,Trueba, P. E., et al (1991)
not target the prevention of depression Peer support telephone dyads for elderly women: was As psychiatry is faced with a Govern-
but examined therapeutic or preventive this the wrong intervention? American Journal of ment currently determined to medicalise
strategies for other primary disorders and ,
Community Psychology, 19, 53^74. as far as possible the growing problem of
used depression scores as secondary out- ¤
Jane -Llopis, E., Hosman, C., Jenkins, R., et al (2003)
Jane-Llopis, violence in our society, it is essential that
come measures. For example, Bisson et al Predictors of efficacy in depression prevention psychiatric journals present statistics on
programmes. Meta analysis. British Journal of Psychiatry,
(1997) studied the efficacy of psychological 183, 384^397.
this subject in a meaningful fashion. Dr
debriefing on the development of post- Salib’s paper, although not specifically
Lewinsohn, P. M., Antonuccio, D.O., Steinmetz, J. L.,
traumatic stress disorder (PTSD) in victims about trends in homicide over time, pre-
et al (1984) The Coping with Depression Course. A
of acute burn traumas. They showed that Psychoeducational Intervention for Unipolar Depression.
Depression. sents misleading data on this subject, which
psychological debriefing may even worsen Eugene, OR: Castalia Publishing Company. are neither helpful nor informative to the
the long-term course of burn victims. But Seligman, M. E. P., Schulman, P., DeRubeis, R. J., et wider debate on violence in society.
while psychological debriefing may have anxiety.
al (1999) The prevention of depression and anxiety.
been mistakenly considered helpful for pre- Treatment,
Prevention & Treatment, 2, article 8. Richards, P. (1999) Homicide Statistics (Research paper
no. 99/56). London: House of Commons Library.
venting PTSD in the past, no reasonable
C. Kuehner Central Institute of Mental Health, Simmons, J. & Dodd, T. (2003) Crime in England and
therapist or researcher has ever claimed
PO Box 122120, 68072 Mannheim,Germany Wales, 2002/2003 (Home Office Statistical Bulletin
that massive emotional confrontation 1358-510X, 07/03). London: Home Office Research
would represent a promising strategy for Development and Statistics Directorate.
depression or depression prevention. Salib, E. O. (2003) Effect of 11 September 2001 on
Second, the coding of respective meth- Homicide data suicide and homicide in England and Wales. British
ods looks rather inconsistent, and I wonder I am writing to query the homicide statis- ,
Journal of Psychiatry, 183, 207^212.
how the authors were able to reach such a tics quoted by Dr Salib (2003). The figures
high interrater reliability across codes. For he quotes for total annual homicides sug- R. P. Rowlands Chesterfield Community Mental
HealthTeam, 42 St Mary’s Gate,Chesterfield S41
example, the psychological debriefing meth- gest a fall in homicide between 1979 and
od used by Bisson et al (1997) was coded as 2001. The source for his figures is quoted
‘behavioural, cognitive and educational’ as the Office for National Statistics
(p. 389), while the code ‘cognitive’ was (ONS).
missing for Seligman et al’s (1999) interven- Homicide statistics are easily available Author’s reply: Dr Rowlands raises an
tion based on cognitive therapy. Similarly, through the website of the ONS and important question, triggered by homicide
four research groups using similar variants from various other sources, including data in my recent paper on the effect of
C O R R E S P ON D E N C E
September 11 on suicide and homicide in Mental health and psychiatric factors are very similar: 4.48 and 4.83,
England and Wales. He argues that when research in Brazil respectively (data from ISI, reported on
tackling violence in our society, the current Saxena et al (2003) have shown the under- http://in-cities.com/countries).
Government may plan services on the basis representation of low and middle-income Although health problems in develop-
of information that is misleading and flawed. countries on the editorial boards of ten ing countries account for over 90% of the
The data used in my paper – in excess leading psychiatric journals, based on a world’s potential life-years lost, only 5%
of 130 000 unnatural deaths (E950–959 World Health Organization report. Horton of global health research funds are devoted
and E980–989, excluding E988.8) – were (2003), Editor of The Lancet, has presented
to these problems (Mari et al, 1997). The
obtained from the ONS in 2002 then some evidence of publication bias against investment channelled to postgraduate and
updated in 2003; 7400 of these deaths were diseases of poverty studied in developing human resource educational programmes
classed by the ONS as manslaughter and countries. Wilkinson (2003), formerly Edi- in Brazil has assured the country a modest
unlawful killing (homicide; ICD–9 E969). tor of the British Journal of Psychiatry,
Psychiatry, but continuous contribution to the world-
It was clearly pointed out in my paper has suggested that the absence of represen- wide production of knowledge in health.
that routinely collected data was a major tation on the Editorial Board does not ne- It is expected that the quality of the scienti-
limitation of the study, but I had to accept cessarily bias an editor’s decision-making. fic production of countries such as Brazil
the nationally collected data from ONS as However, Catapano & Castle (2003) have will influence editors’ decision-making
reliable and as complete as possible. It shown that research papers from develop- and overcome eventual ‘institutional
should be pointed out that before 1993, ing countries represent a very small propor- racism’ (Horton, 2003).
ONS data were based on year of registra- tion of the publications (52%) in (5
tion of death but the data that were actually important psychiatric journals, which has Declaration of interest
used in the analysis relating to September remained the same for 10 years. We argue J.J.M. and E.C.M. are Editors and R.A.B. is
11 related to the year when suicide and that Brazil, a middle-income country, is an Associate editor of Revista Brasileira de
homicide occurred. progressively improving its scientific pro- Psiquiatria.
The paper made no reference whatso- duction and reaching the standards of
ever, implicitly or explicitly, to homicide high-income countries.
trends in England and Wales since 1979. We have assessed the mental health
Catapano, L. A., Castle, D. J. (2003) How
The only comment about trends in homicide Lancet, 361,
international are psychiatry journals? Lancet, 361, 2087.
scientific production of Brazilian postgrad-
was made in relation to seasonal variations uate programmes between 1998 and 2002
to show that the reduction in homicide noted using a Brazilian Ministry of Education
Horton, R. (2003) Medical journals: evidence of bias
poverty. Lancet, 361,
against the diseases of poverty Lancet, 361, 712^713.
after August was not related to the events database. The eight doctoral programmes
of September 11 but merely represented in psychiatry and psychobiology, all in state
some seasonal pattern. The higher homicide Mari, J. J., Lozano, J. M. & Duley, L. (1997) Erasing the
institutions, have awarded 183 PhDs and BMJ, 314,
global divide in health research. BMJ, 314, 390.
figures that Dr Rowlands quoted may have this has resulted in publication of 1664
been, as he rightly pointed out, the result of scientific articles in journals; 605 of these
Saxena, S., Levav, I., Maulik, P., et al (2003) How
notification of deaths that actually occurred in journals indexed by the Institute of international are the editorial boards of leading
in earlier years. Scientific Information (ISI). The produc- Lancet, 361,
psychiatry journals? Lancet, 361, 609.
Dr Rowlands has used the paper to make tion of ISI-indexed papers doubled in this
a political point about ‘a Government cur- 5-year-period. The mean impact factor of Wilkinson, G. (2003) How international are the
rently determined to medicalise violence’. I the ISI-indexed journals where articles Lancet,
editorial boards of leading psychiatry journals? Lancet,
fail to see the relevance of his otherwise valid were published was 1.82 (range 0.01–
comment to this paper, the first and so far 29.51); 64% were published in journals
the only available literature on the effect of with an impact factor 41. The number
September 11 on suicide and homicide in of Brazilian articles in psychiatry and J. J. Mari, R. A. Bressan, E. C.
countries other than the USA. psychology (442) published between Miguel Department of Psychiatry, Federal
E. Salib Liverpool University, Liverpool, and 1998 and 2003 corresponds to 10% of Sao‹
University of Sao Paulo, Rua Botucatu, 740^ CEP
Hollins Park Hospital,Warrington WA2 8WA,UK France’s (4129) production, but the impact Sao
04023-900, Sao Paulo ^ SP, Brazil
One hundred years ago
The attitude of the legal profession treatment of the medical witness which the assumption that the medical witness
towards the medical profession once characterised cross-examination but must not only be prejudiced but ready to
which at the same time did not prevent give perjured evidence on behalf of the
IN a letter published in THE LANCET of Feb. the most vigorous investigation of the facts, party employing him. There are many
27th, p. 611, Dr. E. MAGENNIS, writing of and he drew attention to the unwarranted who agree with Dr. MAGENNIS, and who
the conduct of barristers in Ireland, de- impertinence, frequently amounting to po- will add that the discourteous treatment
plored the disappearance of that courteous sitive insult, which appears to arise from of the medical witness is not altogether