I understood from the report that good-quality (at least as far as the authors of the report were concerned) could be achieved by a non-randomised design, so long as the study was of sufficient duration and those assessing outcomes were blinded to the status of the participants.
(my bold)
The requirements for good-quality would fairly easily be met by randomized-controlled trials, but depended on some arbitrary, less-likely-to-be-met criteria for any other kind of study. Even a well-designed cohort study wouldn't necessarily meet their criteria. The quality ratings for some of the cohort studies were good, but they were still down-graded, based on the arbitrary choice of whether they identified three confounders, rather than based on pre-established measures of 'good-quality'.
Linda
Last edited:

