Insights From A Digital Health Intervention Trial For Heart Failure
A conversation with Mark Schuuring, MD, Ph.D., FESC, FEACVI, FHFA, University of Twente and Medisch Spectrum Twente
The integration of digital interventions is reshaping how researchers and clinicians approach chronic disease management. The recent ADMINISTER I trial focused on optimizing guideline-directed medical therapy (GDMT) for heart failure patients and offers valuable insights into the challenges and opportunities of this digital health tool integration. Notably, the trial's participant population, predominantly male and averaging 70 years of age, presented unique considerations for trial design and execution.
In this Q&A, University of Twente’s Mark Schuuring, MD, Ph.D., delves into the intricacies of this trial, exploring how the age and gender demographics influenced the study's approach, the surprising adaptability of older patients to digital tools, and the logistical hurdles of implementing a real-world digital intervention.
Understanding the trial participants were mostly male and averaged 70 years in age, what impact did the participant population have on the trial design?
There was a lot of discussion up front, with people saying to start at 18 and limit it at 80 years. And I'm so happy we did not do that. This is an average age and patient sex we see in many heart failure trials, so it is representative of the population. However, there might be underdiagnosis or under-detection in women. On the other hand, they might be more hesitant to participate, but there's no clear answer.
But the clinicians were also hesitant about whether a digital intervention could work for an older population. It actually does work; many older people are supported by family. And that's what I see in ADMINISTER II now and in clinical practice with the telemonitoring program. We have 85-year-old people, even 90, and we were surprised how well many patients handle the smartphone. The more the patients are familiar with the devices, the better trained they get. In some cases, patients write down measurements on paper and then the family helps to put them in a smartphone application.
You might expect these patients would not be comfortable with a personal device to communicate or record data in this way. But you found that people were quite willing and even comfortable with the support of family or caregivers.
In a trial, you want to keep everything as controlled as possible, so there were a lot of discussions about how we could ensure they’re all using the same device. But in the end, we decided not to manage that, because it was impossible to do with different vendors in different hospitals. We started with a patient portal platform. We thought we could do it all with one solution, one EHR, and the patients from the other centers could simply create and use a fake patient number for that solution. But in the end, we opted for a research data platform and had the opportunity to collect all the data from patients directly in the platform. That was easy and safe for study questionnaires and remote measurements from the different centers. But it was a concession. It was a challenge to try to align the different vendors. Using the EHR is a better care pathway. We wanted to fully integrate all of the vendors with the EHR, but it was complicated, and so we opted for the simpler alternative of using the research platform.
Then, we had some patients who had iPhones and some with Android devices. We also had different blood pressure measurement devices. Ideally, all this would have been provided by the research team, but we wanted to have a real-world sample. In the real world, patients have different devices. When needed, we supplied them with blood pressure devices, but they still needed a smartphone or computer to send in the data. There were only two patients who could not participate because they had no smartphone.
About the patient education component, what sort of educational materials were provided to the patient and how did you know it was appropriate for them?
We used information partly based on heartfailurematters.org, a patient website. But instead of telling patients to go to this website, we delivered a tailored version of it to them by email. It was tailored to the group in need of GDMT optimization. That's an unstable phase, and if you get to a stable phase, where you are optimized, it's a different care path.
What were some of the challenges executing the trial?
One of the labor-intensive parts of the study was having a Ph.D. student do all manual notes for 150 patients who presented with a difference in what was prescribed by their doctor and what was recommended by the guideline-directed medical therapy. He made a summary for the clinician to review right before the outpatient clinic visit, like a nudge telling them what to do.
But, of course, you want to automate this. You want interoperability and standardization with the different vendors and integration in the electronic health records. With this manual effort, you can get errors. And if you scale up to over 1,000 patients, you need to automate this process. In many electronic health records, you have medication over here and laboratory values over there. Sometimes we had to collect it from different sites. But that's so outdated.
In the future, when we have many more heart failure patients and not many more staff, there will be this inertia of prescribing these lifesaving drugs; they prevent hospital admissions and early deaths. The technique is so far behind, partly due to political decisions. For example, you want to have your data safely stored. Do you want to be responsible for laboratory values drawn somewhere else, and do you want to have that in your computer system? But for a good implementation of digital working solutions in a scalable operation, if you want to provide thousands of patients with those nudges or electronic health record advice, then it really needs the next step of automation.
What did the principal investigators think about the trial, and was there any skepticism?
It was really a challenge to get everybody aligned. Particularly, this included the discussion with experienced staff about patients who had already been treated for a long time at the outpatient clinic. Some people only wanted to include patients with heart failure who have a new diagnosis. But the guidelines say all patients, and we had this opportunity because of the guidelines updated in 2023. So that's the reason we studied them all.
And there was also a group of patients who were seen both by a clinician and by a heart failure nurse, and that was a pre-specified endpoint. The good thing is this nudge, this digital summary, works for all groups in the pre-specified analysis. Even though they were seen by two clinicians, it still worked. There was still room for improvement. At the beginning, people questioned if patients were seen by both the nurse and a clinician, how much room you could expect for improvement. There's still room. It worked in both groups and also for the hospital type. There were the non-academic hospitals and the referral hospitals, which have much more staff. Both the patient types had room for optimization.
What, if anything, surprised you about the execution or the results of the trial?
It was a short trial, only 12 weeks, but we saw significant impact . And there were clinicians who were really hesitant in the beginning but eventually were overall moderately positive in the net promoter score. They saw the benefit of the intervention.
What then have you learned from the ADMINISTER II trial that you mentioned?
A follow-up to ADMINISTER I, ADMINISTER II will be an outcomes trial with a larger population and with a longer follow-up to see the effect on survival and hospitalization rates. We know that it works for individual pharmacotherapy, but if you use this digital solution and have the nudge in your electronic health records, does it really change the behavior of the clinicians over a longer time? Do we really see that they get more active in optimizing their therapy based on patient questionnaires? Can we get rid of some medications and determine what are the beneficial medications?
We hypothesize that if you have a second opinion in the electronic health record for this population, improvement of GDMT prescription persists. But it's a longer trial, so we need to see whether the results will sustain over the 12 weeks. The effect might be that the clinicians get alert fatigue, and that's a risk if you do a longer intervention.
With digital consults, how could they be successful in other therapeutic areas, specifically other chronic diseases?
Usually in chronic diseases, you have multiple targets and multiple drug classes you want to give. And as clinicians, we should do that, but we also have to run our busy practices. So, it's helpful if you get reminders or nudges. I think there are many groups of patients who could benefit from that. In treatment of coronary artery disease, for instance, we have groups of medications that are provided in the guideline. And optimization of medication is needed if treatment goals are not met. And outside cardiology practice, for instance, it can be helpful for renal failure patients who are on many drug classes. And many groups of clinicians can benefit. You have a group of clinicians who are really aware of updates in guidelines as they occur and others who are a little bit later informed. With this, you get the newest, most up-to-date information and that's helpful in many different groups.
About The Expert:
Mark Schuuring is a non-invasive cardiologist (MD, Ph.D., FESC, FEACVI, FHFA) at the Medical Spectrum Twente and a researcher at the University of Twente (Department of Biomedical Signals and Systems). He is certified in CMR (level 3), TTE, and TOE (Full; EACVI) and is an alumnus of the Postgraduate Course in Heart Failure. As a co-supervisor of multiple Ph.D. candidates, he focuses on eHealth and artificial intelligence in cardiology. Schuuring has published over 90 peer-reviewed articles, including papers in “Nature Medicine” and “European Heart Journal – Digital Health and European Journal of Heart Failure” (H-index 24) and received various grants (Nederlandse Federatie van Universitair Medische Centra, TKI-PPP Health Holland, Hartstichting, Amsterdam UMC, and Stichting Hartcentrum Twente). He is the founder of www.digitalcardiology.net.