Re reading this
retrospective registration of this
feasibility study, the 'secondary study design' is described as a '
randomised controlled trial'.
Interestingly the subjects are their own controls, being assessed over different pre treatment periods without intervention and 'randomised' to eight distinct periods of between seven and twenty four weeks waiting time. However, given there is not a clear pattern to these control periods in practice with seven of the ten subjects having different unevenly spread waiting times and three the same, it would seem that this was not a prospective randomisation but rather a retrospective post hoc analysis. Further with the small size of these eight control groups, seven having only one subject, any inter group comparison will be statistically limited. Consequently I would question how meaningful it is to describe this at present as a randomised trial. It looks like trying to after the fact make use of existing data, using historical accident as an unreliable proxy for prospective randomisation.
Also during the waiting period participants are to be (or were) assessed on a weekly basis. This seems a very time intensive process and one wonders what if any are the implications of so much interaction during a no intervention control period. Time wise how does this compare with the interaction during the active treatment phase, are there any consequences of so much testing and retesting and of subjects becoming so familiar with these subjective assessments? Further with weekly testing the different pre treatment control periods have uncontrolled for differences in total interaction time.
Given the shortness of the waiting times in comparison to the potential duration of the intervention of up to six months, I would also question how meaningful this controlling is. How meaningful is it to compare an average passive wait of thirteen weeks to an up to six month active treatment phase? In practice is it even reasonable to describe this as a control group or as control groups?
Surely for any meaningful conclusions to be made there needs to be a control group with a pretreatment phase of the same duration as the active treatment phase, otherwise how is any effect of the passage time alone to be controlled for? A short pretreatment control followed by a much longer active treatment phase undoubtedly would increase the likelihood of any time dependant spontaneous improvement being misinterpreted as a treatment effect.
With the follow up assessments only occurring after four weeks as already pointed out by
@Esther12, the whole time scale of the project feels very messy and unthought out. For symmetry alone surely there needs to be a six month follow up?
In relation to the outcomes, the primary measure and two of the secondary measures are subjective self ratings, but to be fair the third secondary measure 'school presence' has the potential to be objective, though unfortunately they do not indicate how it is to be (or was) recorded.
Increasingly I am wondering if this is a retrospective recasting of analysis of previous and/or ongoing intervention as an experimental design. So are we not only are we looking at retrospective registration, but also retrospective experimental design? Do we have here not an experimental design, but clinical evaluation of existing treatment? There is nothing wrong with doing this as long as the researchers are open and honest about what they are doing.
Presumably the intention is to use this study as the basis for the application for funding for a full prospective study. Given the confusion in this 'feasibility study' and the possible lack of openness in the design process, the prospective study should it be realised needs very very careful scrutiny. However given the lack of objective measures and the historical design failures and ethical failures of researchers advocating this approach I do not have any confidence that a meaningful study will emerge.
