- DEMENTIA RESEARCHER - https://www.dementiaresearcher.nihr.ac.uk -

Blog – Non-pharmacological interventions

I have a passion for neuropsychological rehabilitation and behavioural therapies for people with dementia because I think they can significantly improve quality of life and health outcomes. I conducted my first interventional study ten years ago. Since then, I have developed and evaluated a suit of interventions and assistive technologies and led several single cases studies and clinical trials. The more I do this type of research the more I enjoy it. But also, the more wary and concerned I become about appropriate risk of bias control. Non-pharmacological studies have historically been of poorer quality than pharmacological ones because variables influencing behaviour are more difficult to control than variables influencing the response to a drug. This, obviously, does not help to inform evidence-based decision and convince commissioners to fund rehabilitation services. However, this disadvantage should not be an excuse, but a motivation, to keep raising the quality bar. That´s why I thought I would share some basic tips for newcomers to the field of non-pharmacological interventions. So you can get off to a good start:.

  1. Involve an experienced clinician in the design of your non-pharmacological study as early as possible. There are important things that you learn in your clinical practice that you cannot learn by reading academic papers. Many studies go in the wrong direction because they lack significant input from front-line clinicians (or their advice is unheard).
  2. If you are running a clinical trial, involve a clinical trials unit with experience in non-pharmacological treatments from the start. If you are conducting a single-case study, involve an expert in single-case experimental designs (SCED). Good quality SCED are complex. They are not simply pre and post intervention studies, which are subjected to many biases such as spontaneous improvement, placebo effects, regression to the mean, systematic bias, practice effects and rater influence on self-reported measures, to name a few.
  3. If you are a clinician, be very careful with the “I know this works” beliefs. Clinical observations can be misleading and become particularly dangerous for your research when they turn into faith. You should never be more concerned with proving your hypothesis right that with actually testing it in the search for true. Faith (which is distinct from passion) should be left out the lab.
  4. Outcome measure selection. It only makes sense that you pick an outcome measure relevant to the change that you expect your intervention to produce. For that, you need to have a very clear idea of what you are measuring. This becomes particularly challenging when what you are testing are functional outcomes and not scores. This often is the case in interventions where we aim to modify behaviour (here it comes into play your ability to operationalise variables). When what you aim to measure are scores, you should consider whether the measure you are using is valid (this is, whether it is measuring what you want to measure), how robust it is to practice effects (ideally you want one that shows minimal benefits of practice) and adapted to the culture of the individual. You should also find a measure that is sensitive enough (can it detect small changes?) and reliable. Test-retest reliability is the most important one in this type of studies.
  5. Who conducts the post-intervention assessment? Since participants and raters in a non-pharmacological intervention cannot be blind to the treatment administered, we need to ensure as much bias control as possible by other means. It’s well known that participants tend to adopt a generous attitude towards the therapist they have bonded with. This leads to “grade inflation” in self-reported measures. That is why it needs to be a person different from the one that delivered the intervention who collects the post-intervention measures. Or they may be collected in a way that does not involve a human component (e.g. self-completed online).
  6. Use standard guidelines to get you started. There are several tools to help you plan for your non-pharmacological study. If you are planning a clinical trial make sure you use the TIDieR [1] checklist to describe your intervention so it can be replicable. Also use the CONSORT Statement for Randomized Trials of Nonpharmacologic Treatment [2]. Although they are not study-planning tools, these two reporting guidelines can help you prepare. If you are conducting a SCED, you have the excellent RoBiNT [3] scale to guide you in your design. I would also recommend Krasny-Pacini and Evan´s practical guide [4] on the design of SCED to assess intervention effectiveness in rehabilitation, and Tate and Perdices [5]´ book on planning, conduct, analysis and reporting of SCED.

For me as a clinician, there is an intrinsic gravitating force pulling me towards investigating non-pharmacological approaches to treat people with dementia. Only by producing good quality evidence can these therapies make headway and, maybe one day, become the norm and not the exception in the dementia care pathway.


Dr Aida Suarez-Gonzalez

Dr Aida Suarez-Gonzalez

Author

Dr Aida Suarez-Gonzalez [6]is a Clinical Neuropsychologist and Senior Researcher at the Dementia Research Centre, UCL Institute of Neurology at Queen Square [7]. With many years clinical experience working in Spain, Aida now investigating non-pharmacological interventions, services and assistive technologies to support people living well with dementia – this work has included creation of the ReadClear App [8] to support reading for people with posterior cortical atrophy (PCA).

Follow @Aida_Suarez_ [9]