----- Original Message -----
From: "Doug Altman" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, January 14, 2002 9:13 PM
Subject: The CONSORT statement
There was discussion on allstat in November regarding the CONSORT
statement, a set of recommendations for reports of randomised trials. This
response is from the coordinators of the CONSORT group.
The revised CONSORT statement was published in April along with a long
explanatory document. The statement comprises a checklist of 22 items and a
recommended flow diagram. References are given at the end of this email.
These papers are freely available as pdf files from
www.consort-statement.org
1 Not everyone is on allstat. The stated belief that posting comments on
allstat automatically means that the CONSORT group have seen them was
incorrect. Fortunately, the exchanges have been brought to our notice.
2 CONSORT covers matters that relate specifically to randomised trials.
Issues that apply to all research studies are not covered - see also below.
3 Allan Reese wrote:
"There is no mention of validity. The concentration is entirely on the
following of protocol for making measurements, with an assumption that the
correct and useful measurements have been specified."
- First, the word validity has many meanings and we are not exactly sure
what the criticism is.
- Second, the word validity appears 21 times in the long document and is
also in the CONSORT checklist, in each case qualified as internal or
external validity (these terms are defined in the glossary).
- Third, item 6 on the checklist includes the following explicit mention of
the quality of measurements: "When applicable, any methods used to enhance
the quality of measurements (e.g., multiple observations, training of
assessors)."
Whatever was meant, it is clearly incorrect to say that there is no mention
of validity.
4 There are other related initiatives designed to try to improve the
quality of reporting of other types of study. Most recently, the STARD
group have posted advanced draft guidelines for reporting evaluations of
diagnostic tests (see www.consort-statement.org/stardstatement.htm). Also,
there is indeed a group beginning to look at observational studies cohort,
case-control and cross-sectional in epidemiology and elsewhere. (This was
the topic of the original query posted to allstat.)
5 In his 'closure' email of 7 November Allan Reese gives a list of things
that the CONSORT group might consider. We list these with our brief
responses:
* explicit mention of validity, in planning and interpretation
As noted already, we already discuss internal and external validity at some
length. We also discuss the issue of measurement quality.
* clarify multivariate vs multiple observations
Multiplicity is a complex issue. Paul Seed correctly interprets the text as
referring to multiple measurements; we agree with him that it is clear that
the quoted passage does not relate to multiple outcomes. There are several
other places where multiplicity is mentioned, including two in the
checklist itself (it was not included in the original 1996 statement).
* require reporting of software used, with references and, if possible,
exact instructions (commands, code) for the analysis
We see merit in the suggestion of specifying software and will discuss at
our next meeting (probably later this year). We would not expect to see
exact code in published papers.
* warning that the guidelines apply to a routine situation, but researchers
should be alert to exceptions
We clarify that the statement refers only to 2 arm parallel randomised
trials (a forthcoming paper will extend to other types of randomised
trials) and note that several issues are not covered.
* an explicit note that this is a minimal guideline, and that medics should
respect other professionals, take advice, and acknowledge this at the
appropriate level in the publication (ie in list of acknowledgements up to
joint authorship).
We agree. We say " Many items not explicitly mentioned in CONSORT should
also be included in a report, such as information about approval by an
ethics committee, obtaining of informed consent from participants,
existence of a data safety and monitoring committee, and sources of
funding. In addition, other aspects of a trial should be properly reported,
such as information pertinent to cost-effectiveness analysis and
quality-of-life assessments."
CONSORT is not fixed in stone and we regularly discuss suggestions for
additions and clarifications. Such suggestions ought to come directly to
the CONSORT group. This can be done via the web page.
Doug Altman, also on behalf of David Moher and Ken Schulz
2001 CONSORT references:
Moher D, Schulz KF, Altman DG for the CONSORT Group. The CONSORT statement:
revised recommendations for improving the quality of reports of
parallel-group randomised trials. Lancet 2001;357:1191-1194 and JAMA
2001;285:1987-1991 and Annals of Internal Medicine 2001;134:657-662.
Altman DG, Schulz KF, Moher D, Egger M, Davidoff F, Elbourne D, Gøtzsche
PC, Lang T for the CONSORT Group. The revised CONSORT statement for
reporting randomized trials: explanation and elaboration. Annals of
Internal Medicine 2001;134:663-694.
_____________________________________________________
Douglas Altman
Professor of Statistics in Medicine
ICRF Medical Statistics Group
Centre for Statistics in Medicine
Institute of Health Sciences
Old Road, Headington
Oxford OX3 7LF, UK
email: [log in to unmask]
Tel: 01865 226799
Fax: 01865 226962
www: http://www.ihs.ox.ac.uk/csm/
|