Skip to content

eCohorts and the future of epidemiology

March 29, 2013

mobile smartphoneReporter Ron Winslow posted an article in the March 18, 2013 issue of The Wall Street Journal about a privately-funded University of California, San Francisco (UCSF) effort to track heart disease risk in over 1 million adults using mobile technology.

The project, called “Health eHeart”  is described by Dr. Jeffrey Olgin, UCSF’s chief of cardiology, as “a large-scale digital version of the Framingham Heart Study.”  The researchers plan to engage participants by encouraging them to enter their own data (e.g., brief surveys), be available for digital follow-up, and to use digital apps and sensors to record certain biological measures like blood pressure.

Questions for consideration:

  1. Are “eCohorts” the wave of the future for epidemiology? For what types of research questions are they best suited?
  2. What steps are needed to enable the widespread use of mobile technology for data collection in research studies?
  3. What are the new challenges that arise with the use of “eCohorts”?

We look forward to an engaging discussion on these issues.

Posted by the Epidemiology Branch, NHLBI

15 Comments leave one →
  1. March 30, 2013 6:33 pm

    Thank you for drawing our attention to this study. Before responding to the questions above, I plan to take some time to learn more about the study.

    A quick look just turned up this citation describing some of the apps to be used and data to be collected:
    http://www.informationweek.com/healthcare/mobile-wireless/ucsf-heart-disease-study-to-use-mobile-a/240151945
    It also contains some criticisms by Dr. Kvedar, in Boston, including questions about how the data will be used.

    Perhaps we need more information about the study purposes and study protocol before we can weigh its potential contribution.

  2. Mark Pletcher permalink
    April 1, 2013 11:27 am

    We know there will be formidable challenges to running an effective “eCohort”. One of our goals is to conduct the methodological research that is needed to validate the new measurements and methods that we’ll be using. While we know we’ll be able to do things that traditional in-person cohort studies cannot or have not done, we also need to make sure that we can accomplish the STANDARD tasks (e.g., cohort retainment, outcome ascertainment) just as well or better. We think emerging technologies and EMR are going to make these things possible, but we’ll need to prove this.

    • Richard Fabsitz permalink
      April 3, 2013 9:41 am

      Great point on the need to validate these new measures. I wonder where validation should be done. Is it better to do this in the new “eCohort” that is spread across the country or world or is it better to do it in the existing cohorts where the expensive data collection is already available? I realize that many in the existing cohorts may not be sufficiently equipped (both equipment and tech savvy) to participate in the validation study. What are the strengths and weaknesses of conducting validation studies in each?

      • Mark Pletcher permalink
        April 3, 2013 3:39 pm

        Great question. For validating a Fitbit measure of physical activity, for example, it may be very easy to simply administer a gold standard survey within the Health eHeart Study; for more difficult and involved measurements that have traditionally required a large study site effort (e.g., CVD event ascertainment), it might make sense to do at least some of the validation within existing cohorts.

      • Cashell Jaquish permalink
        April 3, 2013 5:27 pm

        I agree that there may be situations where the “gold standard” has been measured in existing cohorts. However, some of the new mobile apps, sensors and social networks are capturing data that have not been measured or well measured in existing cohorts. In these cases validating in the “eCohort” may be easier.

  3. Monika Safford permalink
    April 1, 2013 4:24 pm

    As appealing as the large numbers of real-world people may be in a study like this, as Mark points out, the challenges are substantial. it is worth pursuing but I think it is unlikely that the same level of rigor can ever be achieved this way as would be achievable in a more traditional setting. There will be oodles of missing data if we rely on patient data entry. The key will be to play to this approach’s strengths that are not available through more traditional epi methods. For example, smart phones could be used to track motion and sedentary time far better than any self-reported measures. One could track time spent in the car, at home, at work, walking, etc. If these data can be merged with other data already used by marketers, such as our purchasing history through credit cards, especially food purchasing data, in addition to health services utilization data, it could be quite powerful.

  4. April 1, 2013 6:17 pm

    A scholarly article on use of smartphones in epidemiology studies: http://www.plosone.org/article/info:doi/10.1371/journal.pone.0006968

    • Cashell Jaquish permalink
      April 2, 2013 4:08 pm

      Thank you for this reference. It made me think about how ecologists have been crowd sourcing their science for decades, even before the term existed. One used to be able to take a “vacation” and pay to help collect data at a remote site, now it is easily done with mobile apps. It is worth thinking about how we as epidemiologists might effectively use “citizen scientists” in our work. Might also help scientists’ public image!

  5. April 1, 2013 6:23 pm

    One challenge will be keeping all members of a cohort on the same technology. Most people have specific needs and reasons for preferring specific hardware/software. If equipment is issued to all members of the cohort as an addition to devices that participants already use, this may result in an overload of devices, and non-use of the study device. If the cohort is open to participants contingent on use of a specific device, a range of selection biases might be introduced. The choice of device is one challenge, keeping all participants on comparable devices is another challenge. It will be interesting to see how the researchers handle this particular issue.

  6. April 13, 2013 2:46 pm

    Regarding Cashell’s comment on “citizen scientists”, AAAS just placed a webinar online addressing this issue, http://event.on24.com/r.htm?e=596419&s=1&k=DCD00E1A9D357179B45C9BD4509EBC4D
    The webinar features three speakers talking about crowd sourcing science.
    As I began listening, I wasn’t sure that I saw the connection to epidemiology, but the first two speakers talked about citizen scientists collecting data, and I could see an application to studies where we might want some environmental measurements that had not been foreseen when developing the study protocol. This might be helpful when there is no money in a budget for additional measurements, however, it does appear that funds are required to engage citizen scientists.
    It is also interesting to think of participants in cohort studies as “citizen scientists”, and think of them under this paradigm. How would we change the way we conduct cohort studies, interpret data, and develop new study questions?

    • Cashell Jaquish permalink
      April 16, 2013 2:57 pm

      I will definitely take a look at this. TEDMED will be a great place to start thinking about and discussing “eCohort and mHealth” ideas. I agree that it is easiest to envision “citizen scientists” in the role of collecting environmental data. However it does not need to be restricted to that. There are areas where we do not have very good data at all as Rich mentioned – real time stress, emotions, perceptions, etc. This can raise novel hypotheses for future research. I appreciate your contributions.

  7. April 13, 2013 2:56 pm

    A couple of comments regarding the above discussion of validating new methods – I’m not sure we should assume that current methods are the gold standard, but instead, hold open the possibility that a new gold standard may emerge.

    Also, new technological developments offer opportunities for new measurement methods as well as new interventions, and each need to be evaluated separately. With regards to measurement and data collection, it is entirely possible that new technology will surpass previous methods in some areas. For example, pen and paper food diaries can be replaced or supplemented by photos of portions and food items. We might also find that younger age groups are less and less willing to use paper forms.

    Data are beginning to accrue from many sources documenting the trade-offs and potential of newly developed technologies. For example, this study just published in JAMIA (J Am Med Inf Assoc), Turner-McGrievy et al., “Comparison of traditional versus mobile app self-monitoring of physical activity and dietary intake among overweight adults participating in an mHealth weight loss program.” jamia.bmj.com/content/20/3/513.abstract

  8. April 15, 2013 8:16 am

    I’m keeping my eye on the #TEDMED tweets (in preparation for going there this Friday!) and saw this post listing the items on the physical exam that can now be done by smart phone. They include:
    • Body analysis using an iHealth Scale.
    • Blood pressure reading using a Withings BP Monitor.
    • Oxygen saturation/pulse measured simultaneously with blood pressure, using an Masimo iSpO2 placed on the left ring finger.
    • Visual acuity via an EyeNetra phone case.
    • Optic disc visualization using a Welch Allyn iExaminer case attached to a PanOptic Ophthalmoscope.
    • Ear drum visualization with a CellScope phone case.
    • Lung function using a SpiroSmart Spirometer app to conduct a respirometer test.
    •Heart electrophysiology using the AliveCor Heart Monitor.
    •Body sounds: A digital stethoscope from ThinkLabs auscultates and amplifies the sounds of a patients lungs and heart.
    • Carotid artery visualization using a Mobisante probe.
    I will definitely take a look at it.

  9. Richard Fabsitz permalink
    April 15, 2013 11:21 am

    I appreciate many of the comments made by Tamar regarding the use of citizen scientists, the potential flaw in assuming exam data are the gold standard, and the update on existing technologies for mobile data collection. I think it would also be worthwhile to think about what other data collections might change the world of epidemiology. Mobile would probably advance epidemiology most in collecting data that are highly variable within an individual and need data collected multiple times (e.g. diet, physical activity) or over an extended period ranging from 24 hours to several days or several weeks or even seasonally over a year. Several come to mind beyond what were described by Tamar. For example, stress, anxiety, or depressed mood generally would help us better understand how these relate to acute and chronic disease. They may also allow inquiries for more specifics when reported. What other measures would expand the questions and hypotheses that could be addressed through mobile data collection? Perhaps readers working in this area will focus on those needs for future technology development.

  10. Gregory Marcus permalink
    April 16, 2013 5:02 pm

    I wanted to reply to Tamar’s important question about keeping all members of a cohort on the same technology. In short, for Health eHeart, we don’t anticipate they will all be on the same technology. We recognize that this will be a challenge. We are designing the survey content to work with as many devices as possible (including smartphones of various types, tablets of various types and personal computers/ laptops of various types…including Apple versus other PCs). For example, our participants currently use a pretty good mix of iPhones and Androids, with a scattering of other smartphones.
    This issue will be most relevant to sensors and mobile apps. While some surveys and mobile technologies will be available to all participants, certain sensor-related studies will be applied to subsets. For example, those who already have a blue-tooth enabled device that can be connected to our infrastructure will have the option of downloading their data- clearly there will be selection bias here, but it is “low-hanging fruit.” Participants may be incentivized to “connect” given the opportunity to store/ visualize all of their medical/ activity data in one place. Another approach will be for us to prospectively identify and then invite participants with particular characteristics (such as using Alivecor in a group with paroxysmal atrial fibrillation or a chest impedance sensor in a group with known heart failure). One of the goals of our study is to create a nimble platform that can be used to quickly identify eligible participants and then implement specific interventions/ studies.

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 130 other followers

%d bloggers like this: