Randomisation methods: concealment
BMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7206.375 (Published 07 August 1999) Cite this as: BMJ 1999;319:375
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Fellow triallists and users of trial information will both take to
heart the clear message from Torgerson and Roberts' article. Paper
envelope randomisation does not guarantee unbiassed allocation, as it is
insecure.
They recommend that all trials should therefore randomise centrally,
at a distance from the site of recruitment. Although central or distant
randomisation is feasible in all RCT's in developed countries, only the
largest multicentre trials in developed countries are using it, and these
are usually co-ordinated from a developed country centre. For us, the cost
of either a computer or a manned 24 hour randomisation desk is
prohibitive. Not only this, but neither telephone line nor mobile phone
access cannot be guaranteed at all times, or in all sites. Does this mean
that trustworthy RCT's are a luxury to be restricted to urban centres and
developed countries?
I hope not. Central randomisation secures unbiassed allocation
simply because it ensures that details of the recruited subject are
irreversably and indelibly entered into the trial record BEFORE an
allocation is given. Paper envelopes dont offer any easy way of achieving
this, but modern palm sized computers, such as the Psion, might. It seems
to me that it would be no great challenge to programme these palmtops with
a randomisation algorithm which was triggerred only after irreversible
entry, via the keyboard, of essential patient details. I guess that this
programme and the data on entered patients could be securely " locked"
away, given that local clinicians are unlikely to be master hackers.
Palmtop computers have built in modems, and so their data could be
remotely uploaded to a trial co-ordinating centre by telephone, where and
when one is available, at suitable intervals. If no phone is available,
the supervisor, at regular visits, could simply transfer information from
the memory to another computer. Reliability and battery life are key
considerations, but these are improving all the time. A further
consideration is cost, but the cost of these computers would be small in
comparison with other costs of a randomised trial.
The randomisation algorithm could be as sophisticated as needed
including stratified randomisation or minimisation. The computers at each
study site, (or held by each investigator recruiting patients) could also
be remotely reprogrammed if need be.
I have scanned Medline and the Cochrane methodology database, using
Mesh terms for randomisation and for computers, but can find no mention of
a portable computer based system for decentralised randomisation. I would
be very interested to hear from anyone who has seen such a system, or is
contemplating developments on these lines.
Merrick Zwarenstein,
Cape Town,
South Africa
Competing interests: No competing interests
Distant Randomisation
Torgerson and Roberts1 rightly emphasise the joint importance of
randomisation and concealment (blinding) as substantial factors in
avoiding bias in controlled
clinical trials. They discuss the use of sealed envelopes containing the
treatment allocations and comment on the breakdown of blinding (and hence
randomisation) due to unsecured codes, poor labelling of medication or a
knowledge of the block size (if blocking is used). These are potential
weaknesses but not a 'local' vs 'distant' randomisation issue.
Even with distant randomisation it will still be important for local
access to the treatment allocation to be available for emergency use so
that individual (securely) sealed envelopes may still be necessary.
Medication will still need to be pre-packed and identified only by a code
so that poor labelling is just as much an issue as with local
randomisation. The block size should, ideally, not be known by the
clinical personnel (for the reasons cited by Drs Torgerson and Roberts)
but the potential for unblinding if it is known are the same for local and
distant randomisation.
There are other advantages to distant randomisation, particularly in
multicentre studies (here we usually refer to 'central randomisation').
We are able to easily keep an up-to-date log of recruitment across all
centres; we can introduce a secondary check of essential inclusion and
exclusion criteria; we can more reliably use methods such as
stratification2; and we are able to use various methods of adaptive
randomisation2 such as minimisation3. Hence, in
multicentre studies, I support the message in favour of keeping
randomisation distant from the investigators but would warn that this will
not, in itself, ensure that the goal of double blind treatment allocation
has not been compromised.
1 Torgerson DT and Roberts C. Randomisation methods: concealment.
BMJ 1999; 319:375-376.
2 Day S. Dictionary for Clinical Trials. Chichester: John Wiley,
1999.
3 Treasure T and MacRae KD. Minimisation: the platinum standard
for trials? BMJ 1998; 317:362-363.
SIMON DAY
Manager, Clinical Biometrics
Leo Pharmaceuticals
Longwick Road, Princes Risborough,
Buckinghamshire. HP27 9RR.
Competing interests: No competing interests