Letters Section Editor: Robert M. Golub,
MD, Senior Editor.
To the Editor: Dr King and colleagues1 concluded that there is little evidence that preferences
affect trial validity. However, as these were observational comparisons, any
preference effects in these trials may have been confounded by characteristics
associated with preference. We believe that a better approach to observing
the role of patient preferences on study outcomes is to accept that a significant
proportion of patients who consent to randomization will have a treatment
preference, particularly for the novel therapy.2,3 We
ask patients’ preferences before randomization and then randomize all
consenting patients irrespective of their baseline preferences. In this way
we can examine within the safety of a completely randomized design whether
there is an interaction between baseline preference and outcome. We recommend
that preferences be measured at baseline before random allocation in all studies
where preference may affect the outcome.
Torgerson D, Moffett JK. Patient Preference and Validity of Randomized Controlled Trials. JAMA. 2005;294(1):41–42. doi:10.1001/jama.294.1.41-b