Firstly, let’s talk about what this is.
This is a blog – not an academic or research paper, not a product analysis, and not a study on healthcare.
It describes my experiences as I have pieced together one approach to connected health, and how I went about that.
With that disclaimer, here we go.
One of the big themes in healthcare over the last year has been the concept of “Connected Health” and the “ePatient“, and considerable hope is pinned on the idea that if patients are more active (and discerning) in the monitoring and maintenance of their own health, and use of healthcare services, population health will improve and healthcare costs will drop.
The premise being that autonomy and control will lead to better health outcomes at a lower cost. It all happens at the corner of individual responsibility and public health.
There is a lot of talk in the industry about a patient driven revolution, one that acknowledges that patients understand the impact of their disease and the associated treatments, that sees an urgent need for clinicians and patients to work in partnership , and accepts a need to challenge the status quo of practices and behaviors. (BMJ 2014;348:g1209). Tessa Richards, blogger at the BMJ, speaks to this with regard to patient data, and the role of the patient as an active participant, rather than just the subject of data within an EHR.
Part of this puzzle of improving healthcare is the concept of patient generated data (PGHD), and the idea that data on vitals, diet, and exercise, immunizations, sleep, and use of medications can be monitored by the individual to guide their own choices, and also be sent to their primary healthcare provider (HCP) – typically the person’s general practitioner. These data would fill in a more complete health picture, and the HCP could monitor and see patterns emerge that allowed lower-cost interventions to prevent or mitigate chronic disease. There could be fewer office visits, lower probability of emergency room (ER) visits, hospitalizations, or readmissions.
The three salient components of this view are:
- Increased autonomy and control lead to improved health choices by the individual
- More data over longer periods in the hands of HCPs lead to more focused and timely interventions, that will lower the use of high cost medical services
- Monitoring of basic health and chronic conditions can enable better and cheaper care
So far, so good.
There are some behaviors that this is likely to give rise to, mostly for the better.
If providers know that their notes are going to be shared by their EHR with the patient’s PHR, they are going to take more care and be more complete than if they think only they will ever read them. This also drives patient compliance since the connected patient will be able to re-read the notes and instructions rather than walk out of the consultation room in a daze and then try to remember all the things the provider told them. The implications for patient safety are also important – being able to actually read the provider’s instructions are a vast improvement over trying to recall them from memory.
If patients think that skipping on walking for a few days or eating five burgers in a week is going to wind up alerting their doctor, they might be motivated to behave in a slightly better fashion.
It isn’t that this knowledge stops unhealthy behavior on both sides, they could simply not report, or think “what the heck”, but it certainly is likely to have an effect to the better, and for many problems, just a small change in behavior will be effective.
On the cost side, it means that instead of only showing up at the provider when there are symptoms, unhealthy behaviors and emerging signs and symptoms can trigger an alert to the provider to intervene. A single provider can handle a great many more patients by exception than in person. Monitoring a hundred patients in this way is far cheaper than seeing a hundred at the office, and far cheaper both in terms of time and the level of intervention.
In practical terms, a nurse practitioner could monitor for values coming across from the PHR to their EHR that exceed upper or lower control parameters of a large number of health metrics, ranging from exercise and sleep to compliance with meds, diet, or blood pressure and glucose measurements carried out by the patient. These alerts can trigger them to look closer at what is going on in the data over time for a specific patient, and either respond directly to the patient with suggestions or make a recommendation for an office appointment. They could also routinely examine individual patient records and reach out to the patients with encouragement or suggestions. Since the data include a large number of population health markers, new discoveries in medicine or changes in protocols could lead to a targeted outreach to patients with new information, suggestions, or closer monitoring.
Worth mentioning is that the connected patient is very important to the industry move towards patient care teams and initiatives like the patient centered medical home.
(My) Technology Approach
Of course there are technical and logistical considerations, such as whether people are able to generate health data without undue complexity and effort, whether they are able to get the relevant data into the hands of the HCP, and whether the HCP has the technology to do something with the data.
As an experiment, I put together components of PGHD to monitor some basic health data of my own. I selected a relatively low-cost approach, and one that at face value I would be likely to be able to sustain over a potentially indefinite period.
Firstly, I experimented with a range of fitness apps on the Android platform, and various Personal Health Record (PHR) applications both on the device and cloud based.
Some I rejected after only a short while due to stability issues or ineffective functionality, and I gradually arrived at some specific requirements based on hands-on experience.
I selected, and then signed up for, a Microsoft HealthVault account as my primary PHR, since this provided a fairly comprehensive set of health records in a free, cloud-based solution.
This choice naturally limited the usefulness of many of the Android apps, but since interoperability with a secure and extensive cloud-based repository is important, the remaining apps are more realistic than ones that are standalone and can only store health data on the device.
Health records span an enormous field, from diet, through exercise, to medical records such as conditions, medications, and labs.
To give you an idea of the breadth of the health records that can be stored to HealthVault, here is a snapshot of the fields that can be shared from HealthVault and a carer or an HCP. This is but a small drop in the ocean of data that could be collected on a person’s health, and a truly quantified self would be an epic undertaking, far beyond what we can currently achieve with any sort of scalability.
Figure 1. HealthVault Sharable Data
My next step was to transition from the somewhat inaccurate (but free) step-counters and fitness apps that I could download onto my smartphone, and buy a FitBit Flex.
The Flex was chosen according to:
- Interoperability with HealthVault
- Features and reputation
The FitBit Flex tracks some things automatically, and more things manually.
|Very Active Minutes
Sleep activity is one of those that is partly automated and partly manual input, and here’s how it works.
Since the Flex can’t tell if you are awake or not, you have to put it into “sleep” mode manually, and then take it out of sleep mode again when you wake up. This of course leads to some days of missed data because you either forgot to put it into sleep mode or to take it out again. The way in which it is done also presents a few challenges – it works by tapping on it three times in rapid succession, and this is mimicked by some day to day actions like knocking on a door, clapping, or some kitchen activities.
So if you applaud during a show, you have to verify that the Flex doesn’t think you are sleeping.
To get the data into HealthVault requires setting up the FitBit smartphone app and configuring HealthVault to receive data from FitBit.
This process was fairly smooth, but not without its share of oddities, like waiting 24-48 hours for the first upload. Initially I thought I must have misconfigured and wasted a lot of time troubleshooting a working configuration.
Obviously not all fields are interoperable between FitBit and HealthVault, and even within a single concept, not all the field dimensions are interoperable or visible to the user.
For example, sleep and exercise data are transferred but there isn’t a perfect match of fields.
||18 min x 13
||11 min x 2
Some values captured by FitBit, such as the Restless and Awake periods are not being used by HealthVault, while two of the values that HealthVault evidently received don’t appear to match anything that the FitBit app displays. “Settling minutes” is either received or calculated by HealthVault, but doesn’t show up in the FitBit app, and the meaning of “Wake State” is unclear and doesn’t obviously map onto anything in the FitBit app.
||Number of steps
|Very Active Minutes
The exercise interoperability maps slightly better, but also has a field in the FitBit side that doesn’t map to anything on the HealthVault side.
All in all, the average user should be able to navigate and configure this without help, and can get some basic health data uploaded from a device like FitBit to HealthVault.
Medical images are a challenge, and typically if you want to provide images to your provider or upload to HealthVault just for your own record, you need a third application that can convert jpeg or png images to Digital Imaging and Communications in Medicine (DICOM) format. It isn’t as simple as taking a photo of the lump on your hand and uploading. Firstly you need to get the photo as a jpeg (easy with a smartphone), upload to an application such as MIPAV and then navigate around a fairly large number of fields to be filled in that are part of the DICOM metadata standard. Some DICOM apps are somewhat inscrutable and at this point they probably assume you are a radiologist or somebody in the field. Typical data you will need to enter include Date, Patient Name, Description, Study Instance UID, Referring Physician, Study Type, and Body Part. It was obvious that the apps were geared towards practitioners rather than patients doing their own imaging.
In my case it took quite a few attempts, and I had to get HealthVault Support (thanks guys) to help troubleshoot. It turned out that the MIPAV app was incorrectly packing a certain required field with spaces that was meant to be null, and HealthVault was applying the standard rigorously and rejected the image. As a workaround I first converted the image to jpeg-2000, and then into DICOM, and that allowed it to upload.
In terms of general usability and maturity, I would say that image uploads are not yet ready for the average user.
The next challenge is to transfer the data from HealthVault to your healthcare providers or to get your medical images and data from your provider.
For this I set up the HealthVault Message Center, and there are a number of options.
At the lowest level of interoperability, you can simply print out the health record from HealthVault and take it with you to the provider appointment, and then get paper records back from them after the appointment and re-enter the data into HealthVault or scan them and load them into HealthVault– not very satisfactory either in terms of efficacy or security. Losing a piece of paper is a very real possibility for some of us. Scanning them in is fairly simple if you have a scanner, and you can upload them either as a Continuity of Care Document (CCD) or a Continuity of Care Record (CCR), This is a bit of a cheat, since CCR and CCD are actually competing healthcare record standards, and have specific fields and meanings that would not be parsed into computable health data from just uploading a scanned document. However, it is better than leaving the paper lying around in a file folder, and if they are in HealthVault you at least have an opportunity to find them again and nobody else will chance upon them while looking in your filofax for the electric bill.
A second option is to set up a provider or custodian in the Message Center, provide their email address and optional password, and then select which of the data in Figure 1 you wish them to be able to see. This however requires a provider to navigate to the website and to go through the login process. It is more secure, under the patient’s control, but requires a fair amount of effort on the side of the provider, who will need to keep a record of the login details for each ePatient. This is not really a scalable model since from the provider’s side it would require them to keep a record of each patient’s chosen PHR, the login procedure, and login credentials. Not all patients will share the same things, and not all PHRs will have the same fields available or in the same format, so it would be a very complicated world for the provider once significant numbers of patients share in this manner. It is the flip side to the mistaken idea that patient portals are a solution, but in that case each patient would have to keep a record of multiple provider portals, logins, etc. Since the average person has ~4 providers, and patients with chronic illnesses have ~15, portals simply don’t scale well.
The most integrated option I had was to set up a Direct account in HealthVault that gives you a <yourname>@direct.healthvault.com address.
This address only sends between Direct addresses, and is encrypted – so no spam and pretty secure. No paper records lying around, no manual portal-surfing, and no proliferation of user codes and passwords to remember. The patient sees everything in their chosen PHR and the provider sees everything in their EHR.
HealthVault even provides you with a natty little printout or email that you can give to your provider that tells them what to do:
If you are using an electronic health record (EHR) system that is certified for Meaningful Use Stage 2, then your software may be able to generate a CCDA and send it to me using the Direct protocol. (As you may know, Direct is a security-enhanced health messaging protocol designed to help protect health information when it is sent from one computer system to another.) Your EHR software vendor should be able to provide instructions. If you can’t yet send information via Direct, can you give me electronic records another way, such as on a disc? HealthVault accepts structured information in CCDA, CCD, CCR, and BlueButton formats, as well as information in unstructured files such as images, PDFs, and text. You can find more information about HealthVault and how it supports Meaningful Use Stage 2 at http://www.healthvault.com/providers.
Microsoft provides ample educational and instructional materials for providers, such as this overview on sending health information to patients
This option allows health records to be sent securely and effectively from the provider’s EHR to your HealthVault account and vice versa, with no extra work on the providers side, and only one login to your HealthVault account to you as the patient.
So far so good.
Unfortunately for me, this is where the wheels come off because of my four healthcare providers, none are able to use Direct.
- General Practitioner in Colorado: Only has paper health records
- General Practitioner in DC: Has an EHR, but staff don’t know how to get it to work to provide electronic records. (They spent huge money for an EHR but still print records)
- Dentist: has an EHR, but doesn’t have the functionality to work with Direct. Instead sent me my dental images over unencrypted email (!)
- Optometrist: has images on a standalone machine that doesn’t connect to anything, and all other records on paper
So at the moment I have limited ability to do any real work as an ePatient or be part of a meaningful care team with my providers. The technology is ready enough, cheap enough, and usable enough to support at a minimal level, but my providers are just not there yet. The question that occurs to me on many of the #bioethx, #hcldr and #hcsm tweet chats, is that at some point I might start looking for providers that are further along the curve, and are willing and able to connect. From a provider perspective, this might be a competitive advantage issue, and providers that aren’t able to offer ePatients a workable data exchange schema may find that their patient population is dwindling and they are left with high-cost low profit patients.
So what did I learn?
Firstly, you can get basic health info including allergies, insurance details, vitals, and essential fitness data into a handy, secure, and easy to use PHR.
Secondly, if you are conscientious, you can keep track of things like blood pressure, diet, water consumption, alcohol and smoking, etc. in the PHR, but you will have to stick with it and remember to keep inputting the data.
Thirdly, if you have a participating provider, and they are also ahead of the curve, you can start transferring some pretty useful health data to them, and get responses back with useful guidance. This should enable you to have better health, get expert advice, and do so cheaper and more efficiently.
The future looks better though – and probably just in time for Meaningful Use 3: weight, blood pressure, sleep, etc. will be things that can automatically be sent to a provider.
The ability to collect and share health data with a virtual team of providers is a game changer, and is allied to the ability to set personal targets, monitor dietary intake and exercise, collect vitals over time, through the combination of wearable tech and Meaningful Use.
It occurs to me when I wait for my turn in a provider’s waiting room, that the bulk of visits to the provider could be done remotely with these tools, and my provider team need not be in the same practice, town, or even the same country. There is nothing at this point that should stop the ePatient from building a care team themselves that might span the globe, and achieve better healthcare, cheaper. A big artifact of the connected patient may be the mass customization and commoditization of healthcare that many other industries have experienced over the last 30 years.
Perhaps it is time.