This week, Stanford’s MyHeart Counts Cardiovascular Health Study published initial findings from their ResearchKit-based study. And for anyone interested in digital health, particularly research, their findings are a must read.

MyHeart Counts launched with the first round of ResearchKit apps back in 2015. In a paper published in JAMA Cardiology this week, the team behind this study shared some initial findings focused on the feasibility of what was one of the first digital health cohort studies.

Between March and October 2015, 48,968 participants consented to participating in the study through the MyHeart Counts ResearchKit app. It was a group dominated by young mean (mean age 36 with IQR 27-50, 82% male). The group was generally healthy too – only about 7,000 reported having any disease and about 3,000 reported taking a medication.

Among other things, participants were asked to complete an initial 7 day monitoring period where motion was tracked using the M7 coprocessor chip in the iPhone as well as daily surveys and a 6-minute walk test at the end of the 7 day period. Of the nearly 50,000 participants, 9% completed the entire 7 days of motion data collection while a plurality (41%) did at least four days. About 10% of participants did the 6-MWT and ~3% did the Heart Age Questionnaire, which assesses risk factors for heart disease.

It was also noted that engagement with the ResearchKit app dropped of markedly after the initial seven day period. That’s a finding echoed by the initial experience reports from the Asthma Health ResearchKit study.

These findings highlight some of the key challenges and opportunities in digital health research. They were able to recruit a remarkable number of patients in a short period time. However, there is clearly going to be a lot of selection bias at many different levels of participant interaction with the study.

They did nonetheless report some preliminary findings from the MyHeart Counts study. For example, they found that participants were way off when it came to assessing their own risk for cardiovascular disease and estimating their physical activity. That insight could generate some interesting hypotheses for further testing, like how giving people more accurate feedback on their own risk profile affects their behaviors.

Digital health cohort studies are growing rapidly, including the enormous Precision Medicine Initiative set to get underway. Clearly, a focus on engaging more diverse participants and developing strategies to maintain engagement are going to be critical, particularly when working with a “healthy” cohort.

The Framingham Heart Study has been going strong from decades in no small part because the people in the town of Framingham are incredibly engaged as individuals and as a community, sharing a belief that they’re doing something important. Somehow, that sentiment needs to be translated into the digital health studies of the future. Perhaps that will be accomplished by investing more in building digital communities within these large studies or even engaging specific real world communities. Or maybe some participants need to be more engaged in the research process, perhaps by sharing results and highlighting potential benefits. In any case, digital health studies like MyHeart Counts that openly share their experiences, allowing the research community to learn from collective experience, will help us figure out how to do that.

Reference: McConnell et al. Feasibility of Obtaining Measures of Lifestyle From a Smartphone App – The MyHeart Counts Cardiovascular Health Study. JAMA Cardiology. 2016. Link