Let me take you back to the beginning. When I first applied for my NIHR fellowship I had to put down a training plan – and cost for this – in my application. At my consequent interview one of the points made by the reviewers was that I would need more training in trials and mixed methods. To this end I have sought out various training opportunities to equip me with more of these skills. Frequently this has meant attending courses not entirely designed for the likes of speech and language therapists. But (I reflect frequently) this isn’t anything new for an SLT. We are often in the minority and by now I quite enjoy the challenge of trying to apply other perspectives to our work- or tweaking concepts to suit what we do!
To address some of this feedback from the reviewers I recently attended a couple of half day courses designed for people involved in running cancer trials. One of the most important reflections I made during this course was around culture. The course facilitator explained that around 20 years ago there were real difficulties recruiting people to participate in cancer research, Consequently the National Clinical Research Networks were set up. As a consequence of this organisation ( who provide research nurses and so forth to help recruit and collect data) it has become part of routine culture for people to participate in cancer research and around 1 in 6 people with cancer now participate in research. At present we have no such figures for SLT. There are lots of barriers to research in clinical SLT practice. Including culture – perhaps we need to consider how we can address some of these issues. Perhaps we actually need to do more research and more trials in clinical settings to break down this culture.
But culture is not the only barrier. Going on this course challenged my thoughts on what else we could borrow from the cancer trialists. Here are some of my reflections:
- For starters we need quite a large number of fairly homogenous patients to do a proper RCT. Or perhaps not. Perhaps simply using well matched or groups that are evenly heterogenous. Never the less we need quite a few people- perhaps we need to be attempting more multi-site and international collaborations?
- When we choose our comparator condition we might want to borrow from practice in cancer trials. Working with some of the rarer cancer groups does not warrant a “standard practice” comparator. Often there isn’t one that a ) is standard practice and b) is already proven.
- What about outcomes? We need to choose clinically relevant outcomes. Ones that change practice. Ones that convince the skeptics. But sometimes we aren’t able to quickly or easily measure the exact thing we are affecting. In these cases it’s OK to use a related or ‘surrogate’ measure in cancer trials. Perhaps we could fall back on confidence more often? Also of note- when you are piloting a research study (phase 2) you may not use the most meaningful measures but those which give you the information quickly in order to plan for the larger phase 3 RCT. Then use the most meaningful ones in that phase 3 trial.
- The cancer trialists will generally use a phase 1 study to identify maximum dosage- using a 3 by 3 design where 3 people are given a dosage calculated on the minimum required to make a change. If no one has an adverse event then 3 are given a higher does, this goes on until at least two people have adverse events in the group of 3. The level below is then deemed the maximum tolerated dosage. Can we reverse this model for SLT- can we work out the minimum dosage to make a change by reducing the dosage incrementally? We need to know what the minimum dosage is that commissioners should fund. So can we borrow models like this?
- In cancer trials tumour size is often used as an outcome. In these cases the tumour is measured at baseline, then during therapy, after therapy and then some time later. If the tumour shrinks by more than 30% this is deemed a partial response, if it increases by more than 20% this is a progressive disease. If it remains within the 20% or the 30% boundaries this is stable disease. Could we consider some of these percentage thresholds as applicable to our progressive patients? It is ever so difficult to measure if therapy is effective in progressive aphasia and stable disease is difficult to define. Perhaps we don’t need to re-invent the wheel? Can we model their response models?
Sent from my iPhone