Digital tools to make new academic research clinically applicable in day-to-day medical practice
Calculate by QxMD converts recent research publications into practical handheld tools. With over 450 unique calculators and decision support tools, it provides comprehensive and insightful results that support clinical practice.
We recently spoke to Sean Barbour, Medical Lead and Chair of the BC Provincial Renal Agency Glomerulonephritis Committee and GN Registry about his experience in working with us to facilitate the process of getting new knowledge into the hands of front-line clinicians. Without further ado, here’s the conversation we had:
This interview has been edited and condensed for clarity.
Hi Sean, thank you for joining us. Before we begin, can you tell us a bit about who you are and what you do?
My name is Sean Barbour. I’m an assistant professor at UBC in the Division of Nephrology. I have a clinical specialization in glomerular diseases which is a group of auto-immune diseases that primarily target the kidney. And I run a provincial GN programme in this area focusing on clinical patients throughout the province. So I see mostly second opinions for other nephrologists or other rheumatologists in the province.
I also help coordinate health services delivery for this group of patients via the BC provincial renal agency. There’s a GN committee in the renal agency that I chair that is mandated with coordinating health services delivery throughout the province for patients with glomerular diseases and integrating it within the existing framework for health services for things like CKD or end-stage renal disease. And then, I also do clinical research in this area. We do a combination of observational research, prediction modelling and also running clinical trials provincially for patients with glomerular disease.
Right, so you hold several peer-reviewed grants including one from the Canadian Institutes of Health Research (CIHR) to advance care in IgA nephropathy. Could you tell us more about your work there and what’s coming on the road map for advancing care in IgA nephropathy?
Yeah, certainly. So my grants through CIHR have mostly been around one stream of predicting outcome in IgA nephropathy. So IgA nephropathy is one of the types of glomerular diseases and it happens to be the most common type. It’s also one of the most challenging to treat because the risk of the disease progressing is extremely variable. For example, in some areas of the world such as China and Japan and other areas of Eastern and Southern Asia, it’s a very common cause of end-stage kidney disease.
So certainly there is a large portion of IgA nephropathy patients whose disease progresses to result in kidney failure. However, there’s also a significant proportion of patients whose disease does not progress. And they really don’t require aggressive treatments and they can be monitored long-term either through simple blood pressure control or using targeted treatment to lower protein in the urine such as ACE inhibitors or angiotensin receptor blockers.
So the challenge has always been in terms of trying to identify (early in the disease course) what someone’s predicted risk of renal function decline might be. And we had challenges doing that using clinical predictor variables alone as it generally resulted in quite inaccurate predictions.
You could improve this if you choose to follow the patients over prolonged periods of time. For example, if you follow someone for 3, 4 or 5 years and then average their clinical predictor, you can do a better job at predicting outcome. But that sort of defeats the purpose because it’s not all that clinically relevant to predict someone’s outcome after having watched them for so many years. Ideally you’d like to do it earlier around the time of kidney biopsy which is when you’re making the diagnosis of the disease.
Our work that’s been funded through CIHR has been really to derive and then externally validate an international multi-ethnic prediction model that can predict someone’s outcome closer to the time of biopsy. And it’s really been an endeavor of collaboration between investigators throughout the world from China, from Japan, from Europe, North America and South America that really work to put their individual data sets together to create indeed an infrastructure that is large and multi-ethnic with a long period of follow-up duration that allowed us to subsequently derive and then externally validate a prediction model that can do a pretty good job at predicting disease outcome back around the time of kidney biopsy. So early in a patient’s disease course, you can predict whether or not they will or will not have progressive disease.
This work hopefully will be clinically relevant because we only use predictor variables in our models that the average nephrologist should have access to under normal clinical practice including measures of disease severity that you can find on kidney biopsy as well as clinical variables such as blood pressure, measures of protein in the urine and measures of kidney function.
As some readers may know, you recently published that tool (the International IgAN Prediction Tool: https://qxcalc.app.link/igarisk) with us on Calculate by QxMD and have a few more coming soon. Could you tell us more about that experience was like for you?
Absolutely. Developing these kind of prediction models can be quite complex and they’re certainly not that easy to use if you don’t have a way of easily generating a predicted risk for someone who’s in a busy clinical practice and seeing a patient in front of them. Otherwise, this sort of simplifies to an exercise in academic interest that won’t be used in clinical practice if we can’t provide a way for someone to easily input their predictor variables and then have a calculator output that patient’s predicted risk.
We were able to work with QxMD to take our prediction models and embed it within the Calculate by QxMD app. So that a physician can open up the calculator in a very simple way, enter all of the relevant predictor variables and then select the time frame over which they would like to predict the outcome, up to a maximum of 7 years. And then the app does all the work for you and will generate that predicted risk that is specific to the patient’s information that you entered into the app.
So I think it’s actually quite instrumental to taking this project from something of academic interest and actually making it clinically applicable that physicians can use in their everyday practice so they can then turn around and educate their patients around what their risk of progressive disease actually is.
And how did you initially hear of QxMD and why did you want to work with us?
I guess I heard about QxMD a long time ago when I was a resident, when I was working with Daniel Schwartz, who is a nephrologist in the Fraser Health Authority who’s been heavily involved in QxMD [Daniel Schwartz is our co-founder and CEO]. And since then, I’ve used the app for quite a few calculators whether it’s predicting outcome or calculating, for example, fractional excretion of sodium. It’s always been very convenient to use the Calculate by QxMD app to do these calculations in clinical practice.
Many of the nephrology residents and fellows also commonly use the app as well as many of the staff nephrologists for these purposes. So I’ve known about the app for quite some time and I also have a pretty good working relationship with Daniel. So when I was working on this project and realised that really the results were not going to be all that easy to implement in clinical practice if we didn’t embed it within an app calculator. I then approached Daniel and asked him if this would be a project he’d be willing to work with us on to get the prediction model subsequently modified to be included in the Calculate app.
How do you think having the tool go live at the same time as your research paper facilitates the knowledge translation of taking that new theoretical knowledge into clinical practice?
I think it’s actually quite instrumental for knowledge translation. I was able to present the results of the study for the first time at the World Congress of Nephrology in Melbourne. And certainly there was a lot of academic interest in the results but again if you don’t have that capacity to inform your audience of how they can actually use the results in clinical practice, in a way that’s easy, it risks being sidelined as just another interesting academic exercise in terms of how we can risk stratify patients with IgA nephropathy.
However, having the concurrent publication of a mobile app really allows the interested reader then to say, “Okay, I wonder how this prediction model might work?”. For example, in the next patient I see in clinic with IgA nephropathy. And having the advertisement for the app embedded within the publication allows the reader then subsequently go download the app. And then play with the results and see how they actually apply to the next patient they see in clinic because often we get very surprising results. We have a patient that we think might be at quite low risk but actually when you look at the sum total of all the small effects of their predictor variables, they actually have a relatively substantial risk of disease progression.
I think allowing a reader to first learn about the prediction model and become interested academically but subsequently having a tool readily available that they can easily then go implement those results in clinical practice is the key link between an academic exercise and then translating it into clinical practice. So I think without the QxMD app, we really would not have been able to translate these results easily into clinical medicine.
So what kind of impact do you think we might see in healthcare if more researchers worked with platforms like ours to develop easier to use tools for clinicians?
I think that there has been a reasonable amount of work on prediction models in different areas and often the clinical gestalt which is what we often rely on can be inaccurate. And I think it’s because intuitively, physicians tend to consider individual risk factors separately. You know, our brains can’t really integrate multiple risk factors simultaneously. Especially if you have risk factors that are moving in a different direction. One is a higher risk, one is a lower risk of the outcome. You don’t have any idea how those two things might add up together in a patient that has both of those features. So a risk calculator can be very beneficial for that. As long as they’re developed and so long as they’re externally validated in a robust way so that you can have confidence that they can apply to the patients sitting in front of you, having access to technology like QxMD is really what allows you to then take those prediction models and apply them in practice.
It overcomes some of these limitations that we have of using a clinical gestalt that relies quite heavily on considering individual predictor values separately. So I think that the easiest sort of short-term implementation and benefit in clinical practice has to do with providing physicians and patients with a way of accurately determining what an individual patient’s risk of disease progression is.
And then, they can then subsequently have a discussion around that and say to the patient, “What does that mean for you? What does it mean for your long-term kidney health?”. And that might set a framework for a more educated discussion around different treatment options and what the risks and benefits of treatment might be. But at least the patient and the physician will understand what the long-term trajectory might look like in the absence of treatment.
Amazing, thank you again for joining us, Sean. Do you have any final thoughts that you would like to share?
I think there’s a lot of increasing research in novel therapeutics in IgA nephropathy. And I think that recently we’ve all become very cognizant of the risks of traditional treatments like corticosteroids. And certainly, if we want to make educated corticosteroids treatment decision, we really need to understand what the risk of progressive disease is going to be for the patient whom you’re trying to make this decision. Because when we’re dealing with immunosuppressive therapies that have a lot of toxicity, ideally you would like to target those treatments to the patient with the highest risk of progression. Now we can’t do that without having a prediction model in place.
However, moving forward there are multiple clinical trial programs investigating novel therapeutic agents in IgA nephropathy and each one is likely to have a different side effect profile. So for each one of these drugs, there may be a different threshold that we would be willing to treat a patient with depending on their predicted risk of progression. And so I think moving forward one very interesting area of research is going to be mirroring a precision medicine approach that’s been used in other fields to something like IgA nephropathy where we try to identify the individual patient’s risk of progression. And then match that to the side effect profile of the potential treatment to determine a more precision-based treatment approach for that individual patient based on their risk factors for progression and what the risk and benefit of the treatment will be. And certainly that will probably change depending on the type of treatment we’re considering using and how risky that treatment is likely to be. So I think there’s a lot of potential novel research to augment this prediction model as we understand more about newer therapeutic options for IgA nephropathy moving forward.
Calculate by QxMD is a clinical decision support platform that hosts over 400 unique tools encompassing all specialties.
We encourage any researchers who may be interested in converting their research into a practical tool to reach out to us at tools@qxmd.com.
