Modified intention to treat reporting in randomised controlled trials: systematic review
BMJ 2010; 340 doi: https://doi.org/10.1136/bmj.c2697 (Published 14 June 2010) Cite this as: BMJ 2010;340:c2697
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.
Excluding randomised patients violates the intention-to-treat
analysis principle and is usually the source of bias. Abraha and Montedori
identified 6 deviations from the intention-to-treat principle and advocate
that “explicit statements about post-randomisation exclusions should
replace the ambiguous terminology of modified intention to treat”.1 We
share this point of view and add that such statements should include
mention of the potential bias that may be induced by post-randomisation
exclusions. Moreover, we believe that “treatment deviation” (which refers
to patients who did not receive treatment, either because they did not
receive any treatment or because they did not receive a minimum of X
doses) should be specified and that treatment exposure must be related to
blinding. The issue is whether non-exposure depends on the group to which
the patient has been allocated. Obviously, in a trial in which the patient
and/or the care provider are not blinded, we cannot exclude that absence
of treatment exposure is group dependent. In the same way, when we define
treatment exposure as “having received at least X doses”, we cannot
exclude that not taking the last of these X doses can be due to an adverse
event caused by the first dose, for example, which means that the
treatment exposure is group dependent. However, when both the care
provider and patient are blinded, the process that led to total absence of
exposure is completely independent of the allocation result. So the
proportion of non-exposed patients is expected (in the statistical sense)
to be the same in the 2 groups, and the non-exposed patients constitute a
random sample of patients randomized in each group.
Therefore, excluding
these latter patients does not question group comparability nor bias the
study results in anyway. As stated by Ferguson et al., “data on patients
who were prematurely randomised and so did not receive an intervention can
be excluded, as long as allocation to treatment arm cannot influence the
likelihood that patients receive the intervention”2, a principle that has
already been applied.3 We thus suspect that among the 81 trials for which
Abraha et al identified only a “treatment deviation”, some involved total
blinding, and therefore even if non-exposed patients had been excluded (a
piece of information not specified), this does not raise any concern.
1. Abraha I, Montedori A. Modified intention to treat reporting in
randomised controlled trials: systematic review. BMJ. 2010 Jun
14;340:c2697. doi: 10.1136/bmj.c2697.
2. Fergusson D, Aaron SD, Guyatt G, Hébert P. Post-randomization
exclusions: the intention to treat principle and excluding patients from
analysis. BMJ 2002;325:652-4.
3. Chosidow O, Giraudeau B, Cottrell J, Izri A, Hofmann R, Mann SG, et al.
Oral ivermectin versus malathion lotion for difficult-to-treat head lice.
N Engl J Med. 2010 Mar 11;362(10):896-905.
Bruno Giraudeau, Assistant professor of biostatistics. INSERM CIC
202, Tours, France - CHRU de Tours, France; Université François Rabelais
de Tours, France; INSERM U738, Paris, France
giraudeau@med.univ-tours.fr
Philippe Ravaud, Professor of Epidemiology. Assistance Publique des
Hôpitaux de Paris, Centre d’Epidémiologie Clinique, Paris, France;
Université Paris Descartes, Paris, France; INSERM U738, Paris, France
Competing interests:
None declared
Competing interests: No competing interests
Re: Modified intention to treat reporting in randomised controlled trials: systematic review
“Modified” intent-to-treat population is one of the most common manipulations in (predominantly industry-sponsored) clinical trials. I think a sincere way of getting around loss to follow-up/protocol deviations in randomized controlled trials would be to use dichotomous outcomes (dichotomize, if needed) to estimate relevant measures of treatment effect (e.g., RRR) and likely benefit (e.g., NNT) after attributing all loss to follow-up/nonadherent cases as failures. However, a majority of investigators/statisticians from sponsoring companies would be unwilling to do that, as it would reduce the claimed efficacy of their products. Instead, they choose to exclude patients from the analyses in the name of the intent-to-treat principle, losing the benefits accrued through randomization in the process.
Competing interests: No competing interests