Don’t immediately believe the mhealth hype – Dave Clifford

Carrying on the tradition of taking an epic comment and publishing it as a stand-alone post, I’m very happy to feature Dave Clifford’s take on the new mobile health data:

Don't believe the hype machineI care very deeply about numbers and measuring what people are doing in reality versus expectations. I believe that polling is a useful quantitative tool to say something about the universe in the absence of complete information, and I’ve studied polling and behaviors for a bit of my career. I worked at UConn’s polling center in college for a bit, and advised some journalists about how pundits cherry-pick statistics. I worked for DARPA, creating more accurate models for use in experimental biology more akin to those used in physics, and moved to PatientsLikeMe to try and gather incredibly important information on how people with chronic health issues work and live day-to-day outside of clinical settings. Currently, I consult with a number of organizations that seek to use data and information technology to improve healthcare and biological research and translate that into delivery.

Most of the time, I have numbers on the brain. Last week, I noticed a number of people that seemed to be drawing the same conclusions on Twitter regarding mobile health and the most recent Pew study. The conclusion that people seemed to be presenting was this – Susannah’s most recent research at Pew showed that mobile health was a big deal, an idea whose time has come.

In conferences around the country, I’d heard the 2010 numbers and the 2011 numbers be used to show largely the same thing. More and more people were downloading health apps onto their phones and that meant that mobile health app was an open field for developers to move into, since more and more people were going to be using health apps in the coming years. Mobile health was one of the “next big things” in healthcare.

I’ve always been amazed that someone could read the Pew reports on this and come to that conclusion, so I went into the Pew numbers and read them myself and dashed off a quick note to the effect of “No, this isn’t what the numbers show unless I’m reading them wrong.”

Here’s why I think that:

  • Since 2010, Susannah’s reporting has shown that the number of people downloading health apps onto their smartphones has remained roughly constant
  • Since 2010, Susannah’s reporting has shown that the vast majority of what are considered “health apps” are diet, exercise, and weight loss apps.
  • Of the remainder, there are limited cases of disease management apps, drug adherence apps, or personal health record apps being used. We’re talking about handfuls of people out of thousands of people.

And that’s without touching things like “frequency of app use.” I can only speak from my inclinations, but the most used “Health App” on my phone is called “Zombies, Run” and it’s a fantastic app that says “Oh no, zombies! Run!” when I’m listening to music on my phone while I’m running. It’s great and it’s the best thing that I use on my phone when I’m running and I think I’ve used it six times. Six times is a lot of times for a user of any given health app, since most people download a “health” app and don’t use it, and very few people use “health” apps more than once. This is even in the broadest category.

On the flip side, Susannah showed that a large number of people use their smartphone in their healthcare. The majority of smartphone users indicated that they have used their smartphone to search for health information. To me, this seems totally logical.

My smartphone has become a secondary brain for me that I can offload factual memory into. It tells me what streets connect where if I’m in a different city, or who that guy was that sang that song that I like, or what time the football game starts and what time the proper football (soccer) game starts. For the most part, it answers these questions via Google, although some people might use Siri to accomplish the same tasks. I use a very, very powerful app for this called “Browser.”

Sometimes I go to and they say “Would you like to download our app?” and I say “No.” Same with Yelp. Same with urbanpages. I don’t want to download your app. If I download your app my phone will have one more button on it that I don’t need and one more process that I will have to shut down and one more way of getting in to data like phone call usage and where I am in space and I would rather not tell you any of that thank you very much.

Other times I’ll be browsing a website and it will have a lot of flash on it or a bunch of dropdown menus or something else very hard for me to interact with via the old two-finger enlarge and click and I’ll just give up and go somewhere else for the information I need.

Sometimes there will be really great ways for me to do things with my phone that I couldn’t possibly do with any other device and someone will have made a very clever app that I can use that is practically magic. Like SoundHound. SoundHound can listen to a room that I am in, tell me what song I am playing, give me a scroll of the lyrics, karaoke style, and then offer me an opportunity to buy the song. That’s really neat. Similarly, anything with GPS fits into this category. I love the Car2Go app that tells me where there’s an available smart car to rent. GingerIO is particularly savvy about mHealth apps because they are inherently looking at things that a smartphone can passively collect (message frequency, mobility, etc.) rather than forcing a user to interact with an app.

On the other side, there are hundreds of health and medical apps that seek to use things that only a smartphone could do in order to advance care. For example, there are a set of apps that claim to cure disease via sound, light, or vibration from a cell phone. These have not been demonstrated to have any benefit at all. There’s also some health apps that tell you if you’re ready for exercise by counting the number of squats a user can do in a minute via the phone’s accelerometer and then asking the user to measure a resting health rate. These are slightly ambitious in their functional claims.

One of the other interesting findings is the number of users with chronic disease that used a health app on their smartphone. This was only slightly higher among those with chronic disease than those without (21 percent v. 19 percent), but the number of people with chronic disease that went online to look for medical information was lower than those with no chronic conditions (26 percent v. 34 percent.) My guess, if I had to venture one (and without seeing the cross-tabs) is that people with chronic conditions are less likely to have a smart phone to be able to look up health information online (thus the 8 percent gap) but more likely to use a smart phone to manage their wellness (thus the increased use of apps.) It’s certainly a cross-tab to check. Additionally, SMS was a popular means to push or receive information among those with chronic illness, with 10 percent reporting texting related to a health condition compared to 6 percent of those without a chronic condition.

There’s a lesson here for “mHealth,” I think. Or rather, several:

  • Before you make an app, ask yourself “Is the app I want to make essentially a mobile web page? Is the purpose of it to serve information to a consumer on the go?” If yes, there are ways to code your landing page(s) so that they detect the sort of browser being used and provide an optimized experience for a mobile user.
  • What side of the “Digital Divide” do the consumers you want to target live on? The gap is narrowing but it’s still there. Many individuals with chronic disease are older and have lower incomes, both of which are unfortunately correlated with lower digital literacy and lower smartphone adoption.
  • Does your app take advantage of a feature of the smartphone in a compelling way? Does it use the camera or the microphone or GPS?
  • Could your app be an API that runs off a separate, common media (Twitter, email, SMS)? Would it be simpler to integrate it with a separate “workstream” of day-to-day life than force a separate “workstream” on to a consumer?
  • Last, if you’re selling an app that cures a disease or diagnoses a condition from user-entered data that is a very difficult thing to do and seldom supported by rigorous science and the FDA is probably going to scrutinize what it is that you say your app does.

Those are, I think, interesting initial guiding principles for organizations and individuals looking to innovate in mobile health. From this report it’s clear that we’re nowhere near the horizon, but at least we have some stars to guide by.


Dave Clifford is a strategy consultant in Health IT and Science and Tech Policy. He cares very deeply about making data meaningful to learning and practice. He is an alumnus of DARPA and, most recently, PatientsLikeMe.

12 thoughts on “Don’t immediately believe the mhealth hype – Dave Clifford

  1. This is a very well articulated position. I strongly agree. I’ve downloaded and tried dozens of smartphone apps and none have entered my day to day workflow except possibly an GPS exercise tracking app which I use when I forget my custom GPS exercise device. I also like the idea of photographing and crowd labeling food, but it’s not simple enough yet.

    As David articulated, the smartphone seems to work best when it augments workflow that takes place in a larger life context or by making something possible that was impossible or difficult before (spot checking heart rate with the camera without an external device) or adding new capabilities at low friction (RunKeeper’s and iPhone sleep trackers). All of these are “phone as sensor” or “phone as communications endpoint” vs. “phone as platform”. The phone as platform will certainly take off in social amongst the younger set, whether it’s Path or some other new hip mode of interaction. I’m extremely dubious that a consumer + mobile platform will have much impact on health. I think the phone needs to be an extension of a larger system that embraces the larger context – my providers, family, etc. Web search for information is an example of this.

    Frankly, the best use case of the phone for health I’ve seen is longitudinal momentary assessment via SMS on a younger population to track rare disease progress and experiments hand-in-hand with clinicians at the C3N Project ( Your phone doesn’t even need to be smart for that!

  2. Ian,

    Unsurprisingly, I agree with you. One of my larger concerns is the degree to which the Tricorder X-Prize has been hyped in recent days as a smartphone enabled doctor. If we think about what a camera on a smartphone, an accelerometer on a smartphone, and what a microphone on a smartphone could reasonable do we’re limited to things like rate and acceleration as well as frequency of movement, skin flush (a poor proxy for temperature) and respiration rate through acoustic waves/breathing. I’m not sure that the instruments are calibrated finely enough to get a very nuanced view of any of those things and the scales would have to be adequately zeroed between uses. This is, again, presuming that one could create a meaningful inter and intra-individual baseline that wasn’t inherently biased by people doing cute things.

    One thing I would add to this is that I think this has been a very big week for findings in mHealth (although that may be observer bias on my part since I’ve had it on the brain.) I owe debts to:

    1) Rochelle Sharpe at the Washington Post for her work on Apps that make health claims (

    2) The bloggers over at Flurry for their great visualization of adoption of apps:

    3) Susannah Fox, Will Fitzhugh (of 5AM Consulting), and John Wilbanks (of the Kaufmann Foundation) all of whom had to listen to me rant as I turned this

    An additional point that John made in conversation was the development and usage of apps with the express purpose of using them in the short term. If something can be used and is used frequently over a 90 day period for a study, is that good enough? I’d like to extend this thought further and ideate on apps for habit formation or behavior change – Is there a time based trigger that an app could deliver, for example, with enough frequency/annoyance factor to turn something from a “I *should* be doing X but am not” into a “Habit?” Behavioral or mental hygiene could be an interesting place to look for something to change.

  3. You guys are making me smarter by the minute — so happy that Pew Internet’s data can inspire this conversation.

    Another delight is watching the reaction on Twitter, like this from @coding_doc – “Oh no, zombies! Run!” reminds me of Douglas Adams’ crisis inducer's_Guide_to_the_Galaxy#Crisis_Inducer

    (“a watch-like device that can create an artificial crisis situation of selectable severity, in order to sharpen the wits of the user.”)

    Love it!!

  4. I am SO relieved to read something that makes this much sense, particularly coming from somebody with the tech-savvy street cred of a Dave Clifford.

    When I say these things, I know it just sounds like the dull-witted ranting of a middle-aged Luddite raging against the machine. But when Dave says these things, he sounds brilliant!

  5. Here’s a must-read reply to this post by David Doherty:

    I’m taking notes on all of this so I can bring the insights back to my colleagues and improve the way we measure mobile in general, not just in terms of its impact on health & health care. I’ve written before about Pew Internet’s (and every researcher’s) challenge: to write survey questions which make sense to regular people AND to experts. We conduct open research (ie, community peer review) as a core practice in order to constantly improve our methods. I’m honored by every contribution — keep ‘em coming!

  6. Great points Dave. Using the cell phone in a compelling way is the key. I (and a lot of people) like having information tracked about myself, I like using RunKeeper to track my runs. Evernote works great to offload thoughts and ideas to another, more easily searchable service.I’d like a way to measure just about anything about me, calories I’ve consumed (how could this be done transparently?), distance I’ve travelled, calories I’ve burned. I’d like to see sleep patterns, glucose levels and just about anything a physician could test. What I don’t want is to have to manage that data, neither the creation of it, the storage of it, or the accessing of it. None of the creation or storage of the data should get in the way of what I’m already doing, or else the cost in time and attention will likely be to high to make it consistent and useful (how many of us actually do count calories?). It has to be transparent. I just want it available, and I want to be put on a program where I might be able to change some of it and track my progress.

    Simple, yes, but making things simple can be extremely complicated.

  7. BTW, wouldn’t it be great if every meal at chain restaurant had a bar code where you could get nutrition information, then see how much activity you’d need to perform to burn through what you’ve eaten?

  8. This is a very thoughtful article but I think the comparatives require more investigation. I am new to this field, so there might be literature already in this space, but I would like to know how mhealth app adoption compares to video games, business efficiency tools, and other categories of mobile software? Not just adoption and usage but repeat usage (longevity), etc. I mean, this is the essential premise of RWJF’s initiatives in gaming-for-health (

    I think Ian captured some of this well in his comments. The challenge with health app’s vs GPS or Taxi’s is that the smart phone trains us to be ever MORE impatient with ‘getting what i want’ (be it directions, who was in that Seinfeld episode, how much that widget costs at Best Buy, SMSing my kids, etc). How can mhealth applications compete when users are becoming ever more impatient with their phone’s capabilities? At the end of the day, lowering my BP b/c i didn’t eat that burger doesn’t feel half as good as … well eating that burger at In&Out. Yum!

    Anyways, we are certainly at the beginning and I love this space because the iterative cycles and sheer volume of making something stick are inherently much higher than a biotech/pharmaceutical. But I think qualities like community driven results (think online gaming and gaming communities), monetary incentives (money talks), or other positive reward systems are important.

    Pavlov had it right… and I hear the diner bell ringing!

  9. Great observations – and highlights the challenge that Ernst & Young articulated brilliantly earlier this year with their Progressions 2012 Report: The Third Place – Healthcare Everywhere ( ). In effect “everybody in the healthcare business is effectively in the behavior change business.”

    I also struggle with the “all-in” bet by apps that are only released on the most expensive platform (iOS) first. I always ask – and I’m always assured – that an Android version is “in the works,” but more often than not – that’s geek speak for “if the iOS version is wildly successful – we either won’t need to build an Android version – or we’ll easily be able to afford to build an Android version.”

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>