Who considers randomised controlled trials to be inappropriate in the secure populations, and why Pe

1
Who considers randomised controlled trials to be
inappropriate in the secure populations, and
why?  Peter Tyrer Imperial College, London
2
Why are randomised controlled trials so popular
in the evaluation of treatments?

• Because they are the only reliable way of
• (a) ensuring equivalence of populations
• (b) removing bias (when done properly)
• (c ) dissecting out the the effects of the tested
  intervention when many other influences also
  relevant

3
Problems of RCTs in psychosocial interventions

• Much more difficult to identify the precisely
  framed question that is at the heart of the good
  trial and which is much easier to define in
  general medicine
• E.g. does drug x increase the 5-year survival
  rate in cancer y?
• More option for bias (single-blind or masked
  allocation only) and overlap in treatments

4
Example of how these can be overcome

• Keep interventions simple or at least clearly
  separated
• Carry out qualitative studies to identify the
  principal research questions
• Do large studies in several centres that reduce
  bias
• Example of this trial of the care programme
  approach (CPA)

5
Randomised trial of CPA

• (1991) 400 patients on Paddington Vulnerable
  Psychiatric Patients Register randomly assigned
  to CPA (close supervision by a key worker linked
  to a team) in which contact was encouraged or to
  standard care followed up after 18 months

6
Results published in 1995

• Tyrer, P., Morgan, J., Van Horn, E., Jayakody,
  M., Evans, K., Brummell, R., White, T., Baldwin,
  D., Harrison-Read, P. Johnson, T. (1995) A
  randomised controlled study of close monitoring
  of vulnerable psychiatric patients. The Lancet,
  345, 756-759.

7
Does CPA prevent drop-out from services yes?
X28.2, df1 Plt0.005
8
Does CPA prevent admission to hospital? No
X27.6, df1 Plt0.01
9
Importance of result

• It was not what we expected
• It was probably responsible for a large part of
  the bed pressure in psychiatry in the 1990s
• It led to the development of assertive outreach
  and crisis resolution teams (specifically aimed
  to reduce bed use)

10
Main problems with rcts of psychosocial
interventions

• They are usually too small
• They are usually too local
• They are usually mounted by enthusiasts
  investigating their own services or product
  champions rather than independent investigators
• They often clash with vested interests

11
Other problems with rcts in evidence-based
psychiatry

• Over-emphasis in evidence base of pharmacotherapy
  as this is much easier to assess than complex
  interventions
• Distortion created by excessive rigour may
  adversely affect the outcome of the intervention
• Less money (than drug evaluation) means fewer
  studies and less evidence but absence of
  evidence is not evidence of absence

12
Reference

• Mike Slade and Stefan PriebeAre randomised
  controlled trials the only gold that
  glitters?Br. J. Psychiatry, Oct 2001 179 286 -
  287.

13
Why no rcts of note in forensic psychiatry?

• Top-down management
• Security considerations
• Excessive need for control
• Absolute belief that protocols are a substitute
  for evidence
• The principle of randomisation is anathema to
  many staff

14
Security considerations

• In the last resort security considerations are
  paramount but they are also a very useful
  smokescreen for sabotaging a randomised trial
• There is no challenge to the invoking of security
  reasons for a clinical decision even when it is
  made capriciously

15
Fear of disaster

• It is very sad that so many people are afraid to
  put longstanding, but untried, beliefs to the
  test because they fear that a randomised
  evaluation of these will expose them to ridicule,
  censure or dismissal if there is a negative
  outcome. The existence of very similar negative
  outcomes occurring in circumstances of existing
  practice seems to invariably escape attention.
  

16
Answer

• Make sure that rcts are good ones
• Give them the status they require in the
  organisation to be done properly
• Educate all staff into the principles of this
  approach
• (Probably) confine them to secure settings in
  which there is genuine open-mindedness to the
  possibility that not everything in the protocols
  of care is necessary to good practice

17
Anathema of randomisation prison officer at HMP
Whitemoor

• I dont want randomisation to decide who comes
  here. I decide who comes here its my job and it
  needs a lot of thought. If you found it made no
  difference Id be out of a job.