I’ve been banging on about single case experimental research designs (SCED) ever since I studied with Prof Neville Blampied at University of Canterbury. Prof Blampied (now retired) was enthusiastic about this approach because it allows clinicians to scientifically test whether an intervention has an effect in an individual – but he took it further with a very cool graphical analysis that allows multiple cases to be studied and plotted using the modified Brinley Plot (Blampied, 2017), and I’ll be discussing it later in this series. Suffice to say, I love this approach to research because it allows clinicians to study what happens especially when the group of participants might be quite unique so RCTs can’t readily be conducted. For example, people living with CRPS!
Krasny-Pacini & Evans (2018) make the case that SCED are useful when:
1. Evaluating the efficacy of a current intervention for one particular patient in daily clinical practice to provide the best treatment based on evidence rather than clinical impressions;
2. Conducting research in a clinical rehabilitation setting (outside a research team) with a single or few patients;
3. Piloting a novel intervention, or application/modification of a known intervention to an atypical case or other condition/type of patients that the intervention was originally designed for;
4. Investigating which part of an intervention package is effective;
5. working with rare conditions or unusual target of intervention, for which there would never be enough patients for a group study;
6. Impossibility to obtain a homogenous sample of patients for a group study;
7. Time limitation (e.g. a study needing to be completed within 8 months, e.g. for a master degree research. . .) or limited funding not allowing recruitment of a group.
So let’s think of how we might go about doing a single case experiment in the clinic.
First step, we need to think hard about what we want to measure. It’s not likely you’ll find an already-developed measure that is tailored to both the person and the treatment you want to use. There are key characteristics for this measure that you’ll need to consider (these come from the SCRIBE guidelines – see Tate, et al., 2016). You’ll want to look for target behaviours “relevant to the behaviour in question and that best match the intervention as well as accurate in their measurement”; “specific, observable and replicable”; “inter-observe agreement on the target behaviour is needed”.
You’ll also want to think of the burden on the person completing the measures, because mostly these will be carried out intensively over a day/week or even a therapy session.
Some examples, drawn from the Krasny-Pacini & Evans (2018) paper include:
- the number of steps a person does in a day
- time it takes to get dressed
- VAS for pain
- self-rated confidence and satisfaction with an activity
- Goal attainment scale (patient-specific goals rated on a scale between -2 and +2) – this link takes you to a manual for using GAS [click]
- the time a person heads to bed, and the time they wake up and get out of bed
You can choose when to do the measurements, but because one of our aims is to generalise the learning, I think it’s useful to ask the person to complete these daily.
You’ll also need to include a control measure – these are measures that aren’t expected to change as a result of your therapy but are affected by the problem and help to demonstrate that progress is about the therapy and not just natural progression or regression to the mean, or attention etc. For example, if you’re looking at helping someone develop a regular bedtime and wakeup time, you might want to measure the time they have breakfast, or the number of steps they do in a day.
Generalisation measures are really important in rehabilitation because, after all, we hope that what we do in our therapy will have an effect on daily life outside of therapy! These measures should assess the intervention’s effect on ‘untrained’ tasks, for example we could measure self-rated confidence and satisfaction on driving or walking if we’ve been focusing on activity management (pacing). We’d hope that by using pacing and planning, the person would feel more confident to drive places because they have more energy and less pain. It’s not as necessary to take generalisation measures as often as the target behaviour, but that can be an option, alternatively you could measure pre and post – and of course, follow-up.
Procedural data are measures that show when a person implements the intervention, and these show the relationship between the intervention and the target we hope to influence. So, if we’ve used something like a mindfulness exercise before bed, we hope the intevention might reduce worry and the person will wake feeling refreshed, so we’d monitor (a) that they’ve done the mindfulness that night; (b) that they feel less worried in the morning; and (c) that they wake feeling refreshed. All of these can be measured using a simple yes/no (for the mindfulness), and a 0 – 10 numeric rating scale with appropriate anchors (for less worry, and feeling refreshed).
If you’re starting to think what you could measure – try one of these yourself! Start by deciding what you’d like to change, for example, feeling less worried. Decide on the intervention, for example using a mindfulness activity at night. Add in a measure of ‘feeling refreshed’. Keep a notepad by your bed and each night, record whether you did the mindfulness activity, then in the morning record your level of worry 0 = not at all worried, 10 = extremely worried; and record your feeling of refreshment 0 = not at all refreshed, 10 = incredibly refreshed.
If you want to, you can set up a Google Docs form, and graph your results for each day. At the end of each day you could include a note about how stressful your day has been as another measurement to add to the mix.
For patients, using text messaging is really helpful – if you have a clinic SMS service, you could use this to send the text messages to your client and they can text back. Many of the SMS services can automatically record a client’s response, and this makes it easy to monitor their progress (and yours if you want to try it out!).
There are some other designs you can use – and remember I mentioned you’d usually want to record a baseline where you don’t use the intervention. As a start, do this for at least a week/seven days, but you’re looking to establish any patterns so that when you do the intervention you can distinguish between random variations across a week and change that occurs in response to your therapy.
Have a go – and let me know how it works for you!
Blampied, N. M. (2017). Analyzing Therapeutic Change Using Modified Brinley Plots: History, Construction, and Interpretation. Behavior Therapy, 48(1), 115-127. https://doi.org/https://doi.org/10.1016/j.beth.2016.09.002
Krasny-Pacini, A., & Evans, J. (2018). Single-case experimental designs to assess intervention effectiveness in rehabilitation: A practical guide. Annals of Physical & Rehabilitation Medicine, 61(3), 164-179. https://doi.org/10.1016/j.rehab.2017.12.002
Tate, R. L., Perdices, M., Rosenkoetter, U., McDonald, S., Togher, L., Shadish, W., Horner, R., Kratochwill, T., Barlow, D. H., Kazdin, A., Sampson, M., Shamseer, L., & Vohra, S. (2016). The Single-Case Reporting Guideline In BEhavioural Interventions (SCRIBE) 2016: Explanation and elaboration. Archives of Scientific Psychology, 4(1), 10-31. https://doi.org/10.1037/arc0000027
Thanks for this, Bronnie. I have also been banging on N=1 studies for a long time. We actually changed our research component of a post-graduate diploma to N=1 studies but then the students found it very difficult to get it published in a reputable journal [‘too low evidence acc to most reviewers, but I do not think they understand the value of a series of N=1 studies, though]. We will be pressing towards this at the coming teachers meeting of the European IFOMPT congress in September this year