Comparison of descriptions of allocation concealment in trial protocols and the published reports: cohort study

Definitions of methods

Examples

Table A Definitions of methods for allocation concealment and strict criteria compared with loose criteria for assessing them

*Clinician who enrols participants contacts remote centre and obtains treatment assignment.

†Prognostic data for retrospective stratification and minimisation not included.

‡Minimisation was interpreted as being inherently strictly sequentially administered, as allocation of patient will be uniquely determined by how patients’ baseline data correspond to factors for which minimisation is used.

§Treatment allocated either by sequential administration of envelopes containing treatment assignment or by drawing at random.

¶Numbered coded vehicles (implicitly or explicitly described as containing treatment in random order) when no other means of allocation concealment was implied.

**Two studies, that supposedly used numbered coded vehicles for concealment, explicitly described vehicles or their content as distinguishable. These two studies were classified as having inadequate allocation concealment (see table 2).
 
 

Examples of how allocation concealment was assessed for three most common methods

Central randomisation

In an open trial, clinicians contacted the central randomisation office to obtain the treatment allocation of the next patient. The clinicians provided more information on the patient’s prognosis than necessary for stratification.

Comment

We had no guarantee that the central staff person did not use this additional information to alter an unconcealed allocation sequence to "help" the trial to show the desired result; unclear by the strict criteria, adequate by the loose criteria.

Numbered coded vehicles

A study used numbered coded vehicles containing treatment and control treatment in random order, but there was no information on whether the numbering or administration of the vehicles was sequential.

Comment

Here the numbers could have been a random sequence of two numbers (for example, 22212111), meaning that if the blinding was broken for just one patient it would be broken for all. Or if there was no demand for sequential delivery of them to the patients, then known or decipherable block sizes or security envelopes could allow for informed adjustment of the sequence of their delivery; unclear by the strict criteria, adequate by the loose criteria.

Envelopes

An open trial used sealed envelopes to allocate patients to each group.

Comment

No information on whether the envelopes were transparent if held up against strong light, allowing selection of the next patient to be enrolled to have a prognosis that would make the preferred treatment seem superior; or if the sequence of patients was difficult to alter, then the sequence of envelopes could be changed, since these were not described as prenumbered; unclear by the strict criteria; but adequate by the loose criteria.