In this piece, I discuss how the idea around the people-like-me prediction tool (KomPas) came about. KomPas is a tool that makes person-centred predictions regarding the expected (therapy) course for people with Intermittent Claudication. This a story about baseball, Denver, the consultation office, and evidence-based physiotherapy.
In 2012, I worked for eight months as a researcher at the University of Colorado. I had just finished my PhD research and had the opportunity to travel to Denver. After some googling, I found out that Dr Jennifer Stevens-Lapsley’s research group was conducting research in my area of expertise. Two emails and a phone call later, I would start working in her group as an unpaid researcher. Within this group, I had a free role – allowing me to think freely about my field: evidence-based physiotherapy. It may sound crazy that it was only then that I really started thinking about my profession, but in the hustle and bustle of everyday life, it can be difficult to see where you stand or where you want to go. I was even more fortunate that in this research group I met a smart physiotherapy researcher: Andy Kittelson. Andy and I quickly became good friends.
Together, we often talked about the state of physiotherapy. Something that fascinated us was that scientific knowledge is produced on a group level, but ultimately to be applicable, it must be applied by a healthcare professional to a single individual. That reapplication of knowledge is not easy. Around that time, there was also increasing discussion in the medical scientific literature about the value of evidence-based practice. After all, findings from scientific research are not always informative for daily practice. For instance, a study may have been conducted in a completely different population (often in a group of patients without co-morbidities) or in a completely different country with a completely different healthcare system. Moreover, not every study is the same quality (and therefore potentially less valid). Furthermore, keeping up with all the new scientific knowledge has become almost impossible (e.g., about 8000 (!) new scientific articles appear every day). Finally, even when we translate this new knowledge into guidelines, we are only left with advice for the average patient. And just how often do we see average patients in daily practice?
This begs the question – is evidence-based practice a broken and outdated system? At first, that is an easy thought. Surely it can’t be a good idea to make choices for individuals in daily practice solely based on (often inappropriate, simplified, and low-quality) scientific knowledge. Fortunately, evidence-based practice is more than that, something we often seem to forget. In 2005, Dawes and colleagues described that when we apply evidence-based practice, decisions about health are based on the best available, current, valid, and relevant evidence. The authors also indicated that these decisions should be made by those receiving care, informed by explicit knowledge of those providing care, within the context of available resources. In other words, the person receiving care (The Individual Formally Known As Patient) makes the final choice, not the healthcare provider. And that the healthcare professional provides the patient with the best available knowledge (both scientific and clinical reasoning). But providing someone with the best available knowledge is, as I mentioned earlier, the real challenge, especially when this evidenced is constructed outside the context of everyday practice.
Just how do you get the best available knowledge into everyday physiotherapy practice? Andy and I regularly discussed these challenges, and during a Rockies baseball game (Denver’s baseball team), the solution came to mind. Possibly this insight came from the amazing view in the Coors stadium. From the cheapest seats at the RockPile stadium, you can see downtown with a view of the Rocky Mountains in the horizon. A view I wholeheartedly wish for everyone to witness at least once! The solution Andy and I came up with also has to do with views. What if we could support patients and physiotherapists with visuals about the best available scientific knowledge regarding the likely course of treatment?
The way evidence-based practice typically works is that scientific research informs the choice of the best possible treatment. If we adopt the sports analogy, that means looking mainly at previously played games (under controlled conditions); who won from whom. But by no means can we assume that every match is the same. After all, every team has different players with different capabilities. Even the location and layout of stadiums have an influence (and the same goes for physiotherapists and their practices). But it’s also possible that players get injured, requiring a different strategy to be chosen (not everyone responds the same to therapy), and so on. For example, it is important for the Rockies to play at home. Not only do they have the crowd on their side, but they are also used to playing at altitude (Denver is known as the mile-high city) which makes the opponent more likely to get tired. So, the Rockies’ coach and players have little use for global information, but need information that suits their club and their circumstances. How can we harness scientific knowledge in a way that it gives useful information about upcoming games (or, in our case, new patients who inquire about the best course of therapy)?
Andy and I thought that maybe we should switch our focus from comparative research (which treatment works best) to prognostic research (what is the likely course of symptoms). If we know what to expect in terms of outcomes in a single patient, then patients and physiotherapists can monitor the effect of the treatment employed (note: in this situation, you can make excellent use of the knowledge gained from comparative research!). Does the intervention work as expected, better or worse? How do we do this in a person-centred way?
We came up with the idea of translating growth curves like the ones used in the consultation office, to physiotherapy practice. In a consultation office, every new child is weighed and measured against the progression of all children previously measured. Here you have that average again, because now you are comparing one child to all children. But what if instead of looking at the whole group, you only focused on 10 or 20 babies most like the new baby; wouldn’t that be a much more intuitive way of estimating attrition for the new baby? At that time, Prof Stef van Buuren had just shown that this could be a very good approach for babies. Our question – could this also work for physiotherapy practice? In late summer 2012, Andy and I began working to translate this idea into everyday physiotherapy practice. And yes, it works! We call it the people-like-me approach and we have developed a tool for physiotherapists working with people with Intermittent Claudication in collaboration with Chronic CareNet and the Royal Dutch Society for Physical Therapy.
This blog won’t go into detail about the specifics of the prognostic approach but you can find the specifics here. In short, what we found is that our people-like me approach results in both a more accurate and intuitive estimation of a person’s recovery over time than compared with all historical patients. Getting curious? Great! Dr. Anneroos Sinnige wrote her PhD thesis on the development, validation, and implementation of KomPas into daily clinical practice. You can find it here.
In the PREPARE project we are now looking whether artificial intelligence can further optimize our approach. Moreover, through the PREPARE platform we are hoping to share our people-like-me approach with other clinical partners so that other healthcare professionals can make use of this simple yet powerful way to stimulate evidence-based practice. Finally, through PREPARE’s federated learning platform we might be able to create uniform, meaningful database within the field of rehabilitation, enabling us to develop even more of these person-centred prognostic profiles. I can’t wait!
Author: Thomas Hoogeboom, Radboudumc