Google Health: did it have to end this way?

I never fully understood Google Health.  It seems to be a consumer product, inviting you to input your data and track your health, set health goals and so on. Certainly there could be some benefits in this and in the aggregation of information, but the amount of effort maintaining all your records and doing so accurately boggles the mind.  I can’t see myself wanting to do it.

But its real power, surely, would come from the way in which it might operate as a unit patient record, and if it was going to operate as one of those, you needed to get buy in from health systems. Then you could really be cooking with gas with the system inputting data, you doing likewise as well as controlling the permissions that allow people to access your data at different levels of intimacy. With a lot of people included in the system the data would really be powerful when aggregated.

Given that, I thought Google’s strategy would be to build the best consumer app they could but then go hell for leather to get some health systems to interface with them.  Then once some health systems demonstrated the power of the approach, Google would be sitting on top of the incipient standard and – they’d be happy campers, and given that the health systems would not agree to do this without assurances of the portability of data, it’s hard to see how we all wouldn’t be happy campers. I don’t know how much they tried to engage health systems if at all.  They do talk about “adoption among certain groups of users like tech-savvy patients and their caregivers, and more recently fitness and wellness enthusiasts”.

“But” Google reports, “we haven’t found a way to translate that limited usage into widespread adoption in the daily health routines of millions of people,” and so they have just announced the closure of Google Health.

Meanwhile virtually all the governments in the known developed world (I think with the exception of some Northern European countries – mainly Scandinavian ones) have been fluffing around with all singing, all dancing unit patient record architectures which consume billions of dollars but never seem to become operational.  Wouldn’t it make a lot more sense (at least while we wait for these billion dollar monsters to actually deliver, and perhaps instead of them) for governments to outline a set of data integrity requirements (privacy, data ownership, open standards, permitted uses of the data and so on) and then to tender out the provision of a free unit record to Google Health or its competitor Microsoft HealthVault.

By ‘tendering’ the job out, I don’t mean that I expect the government would pay for unit record services from Google or Mircosoft.  The tender would be to ascertain what any firm would be prepared to provide by way of services (including payment to the government, but this would be a secondary matter) for the right to be a preferred supplier of a unit patient record to the health system and to consumers. If, by providing a compelling offering a firm were to win this right it would do the IT to make the platform function and interface with the health system’s IT systems in return for the Government saying to its citizens “Google/Microsoft/AcmeIT is our preferred supplier of unit patient records. You are not required to use them, but if you do, we will commit to ensuring that a progressively larger share of our health system interfaces with your record over the next few years.”

I would have thought that such a prospect would be extremely enticing to a company like Google or Microsoft. Yet if Google tried it, I never found out about it. But I think this is what private public partnerships should be about – and not dodgy finance schemes to get debt of government books.

 

 

This entry was posted in Economics and public policy, Health, IT and Internet, Web and Government 2.0. Bookmark the permalink.
Subscribe
Notify of
guest
18 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Patrick
Patrick
10 years ago

I agree wholeheartedly.

JM
JM
10 years ago

Health is the Gordian Knot of IT. Solve it and you’ll make Bill Gates look like a piker.

There are so many problems in this area that getting something even halfway workable is a huge challenge. Google “Healtheon” and “Jim Clark”

Craig
Craig
10 years ago

Government refused to work with companies like Google due to the interest groups across big pharma and health providers.

Instead they prefer to waste money on systems that will never work.

Health isn’t about preventing illness or even curing it. It is about monetizing peoples’ misery and being seen to be doing something.

shane bonetti
10 years ago

Some time before the last UK election, the Conservative party announced or perhaps mooted as one of their policies the abandonment of all internal NHS IT development projects (because of the lamentable history of spending hundreds of millions of pounds and never getting to the finish line) and moving to use google health. Clearly, that hasn’t happened, but this itself is perhaps no surprise. First, most of the things that the conservatives and their friends the liberal democrates have done were not in the manifestos of either party. Second, the Health Secretary Andrew Lansley has made such an almighty stuff-up of his so-called NHS reforms that some possibly good ideas have been thrown out with the rest of his bilge.

Paul Frijters
Paul Frijters
10 years ago

Nick,

I think you severely underestimate the degree of complexity of actual health systems. This is supposed to be one of my areas of expertise, but at least i know I am only scratching the surface when I say any individual hospital has more than 20 health information systems, and any group of hospitals may well have a 100. There are health IT systems for almost anything imagineable ranging from the accreditation of the nurses to the visits to the pharmacies to the assessments of the GPs. The problems in unifying this are absolutely staggering so I am completely unsurprised Google Health has thrown its hat into the ring.

If you want to have an idea about how the commonwealth government is spending over half a billion to make what it knows to be no more than a small dent into this issue, just look here:
http://www.nehta.gov.au/

Paul Frijters
Paul Frijters
10 years ago

Nick,

the problem with the incrementalist approach is incentives: the only ones to really benefit from the integration of health IT systems are patients. For everyone else it means more work (at least, initially). Why would the directors of hospital X care that their records cannot easily be read by hospital Y? As a result, the health systems have become more and more local in the last 20 years and harder and harder to integrate.

The right comparison is probably the development of scales and weights: individual manufacturers of scales and weights built for local markets, not caring that a weight in one place wasnt easily comparable compatible to a weight in another place. This lead to a great proliferation of weight measurements and thus of scales. It took government intervention to standardise things, enforce education into that standard, buy in some of the agreed upon standards in bulk, and to gradually go towards the system of standardised weights and scales we now have in most countries. Yet still, the movement towards weight standardisation is not quite done, with pounds, stones, and ounces still in existence, 200 years after the french defined the standard kilogram!

I think if you want incrementalist you would need to start with giving the one group that has an incentive to care about this (the patients) all the information that exists about them in order for them to send it to a central google-type depository (or agree to have it sent to something equivalent). That kind of plan already exists, but this is easier said than done. You would be amazed how many records exist of you that are mutually inconsistent and that view you from many different angles.

Paul Frijters
Paul Frijters
10 years ago

Nick,

I think you are missing the main reason why this hasn’t yet happened anywhere (not just in Oz but anywhere): health IT systems are not the result of interactions on an open market with constantly changing partners (i.e. they are not common platform services like the examples you keep coming back to). Health IT systems emerge to help local services do their job.

Let me give you an example and then you tell me how you can reform it so that Google starts to be useful. Let us only look at the point of view of pharmacies and leave out all the other players for the moment.

Pharmacists need records of their clients to claim back money. They also need records to comply with the requirement that they dont proscribe medicines that someone is allergic for. They need basic checks (do not give out a morning after pill to a man!). They need records so they know that a prescription comes from an accredited doctor and that you are their patient. They need records to keep track of the pill manufacturers. They need records to keep track of their own invoices and previous employees. They need inventory records. They need records to keep track of who is an accredited pharmacist. Etc. And that is just pharmacists. You will be on many of those records, with slight variations as to how your name is recorded, your marital status, varying lengths of medical history, your contact details, the details of your current GP, etc. Those records will thus often not even be consistent within a single pharmacy. And you will have visited dozens of pharmacies in your life, in many different countries, each of which will have some record with some version of you on it.

Now, you tell me how to organise it so that the hundreds of pages of information the combined world of pharamcists has on you gets to come to some central website, is transformed in a way that it can be read by other health providers . And bear in mind that most health providers only want to know very particular information about you, i.e. they do not want to read the hundred pages about some foot sore you had when young but only the bit relevant for the thing they are looking at, say, your eye problems).

The market has not even remotely come up with something that helps do this because none of the individual pharmacists have an incentive to care about this (there is no real competition in pharmacy). The same goes for GPs, hospitals, etc.: they do not truly compete for patients as that is not how health is financed, and hence they also mainly care about their internal needs for systems. Even patients have very low-powered incentives to gather the medical information that exists on them. And the basic issue that the group who should care (the patients) are not really customers is not going to go away: there are too many market failures in health for true customer-supplier dynamics to emerge. Hence the state is becoming involved as the representative of the group with most to gain from health data standardisation.

Whilst I thus agree with your skepticism about the ability of governments to come up with standardised health IT systems, I am even more skeptical that there is a natural progression of incremental steps based on self-interest that would get us there. So I throw out the challenge to you to come up with ‘a start’ that a significant group would actually use and that could lead to the dynamic you envisage.

Ed's mum
Ed's mum
10 years ago

It’s very complex. Any change will need agreement from many players and will need to comply with both state and federal legislation. That all takes time and money with any savings in the much longer term.

One solution could be to start with the patient since they have the most to gain. People could compile their own Health Facebook and “allow” access to different health providers with different levels of protection. You could store medication history and test results etc. It would have varying uptake but probably wouldn’t take any longer than a system-led approach. Institutions would modify how they currently deliver information to meet the demand – and it encourage innovative ways to use information stored on different Health Facebook pages.

MikeM
MikeM
10 years ago

I think you guys are barking up the wrong tree. This kind of change has to be driven from the grass roots, and the other day I saw a vestigial sign that this might be happening.

The general practice I go to has kept all its medical records electronically on a server somewhere for a number of years. I don’t always see the same doctor there but whoever I see can always pull up my record and read it in the minutes before I come in. The other day I had an appointment which led to a specialist referral and I was interested to note that at the bottom of the letter it said, “Please send all correspondence via encrypted email (Argus) to mail@xxxx.com.au“.

I should note that the doctor I saw (she has a real doctor with a PhD as well as medical qualifications) was a topnotch touch typist.

I also noticed that the results of a blood test she asked me to take on Friday were available to her when I saw her on Tuesday – a year ago this process would have taken at least a week.

I had a chest CT scan last Thursday and when I went to pick up the stuff on Tuesday it took them half an hour to find it. Now I have to lug it round to the GP, then to the specialist, neither of which will have access to a permanent record of it, despite the fact that what the CT machines spit out is a digital record of a couple of gigabytes, which could be cheaply stored in a central health data repository.

When I went to work for a bank in 1982 it still had people that it employed as messengers. In 2011 the Australian health care system still has people that it employs as messengers, except that it calls them patients.

Paul Frijters
Paul Frijters
10 years ago

MikeM,

your story exactly illustrates many of the reasons this is so hard. Just reflect on some of the elements of your experience:

1. The actual amount of information on you is staggering. Gigabytes for a single scan. Many visits to many different places, each with different protocols, nurses, doctors, etc. Many of whom have little time or incentive to centrally store a record of their interaction with you.

2. Privacy concerns: all this information stored about you and sometimes shared with others. Did you give permission for this? Would you give permission for all health professionals to know everything about your medical record (including any treatments you might have had for embarassing illnesses)? For something like this, privacy laws would require you to be asked everytime you have an interaction whether it is ok for the data to be shared. Just informing you properly about all the ways the information might be used (for your benefit, for the benefit of medical research, etc.) could take quite a while. All this means extra time and work for health professionals, which is neither in the interest of the health professionals nor really in yours (it just means less time helping you).

3. Local needs: your GP has the records of previous visits stored in such a way that they immediately see the information useful to help you. But that is not the format that would also be useful to other health providers. Your scan is not relevant to a pharmacist, who really wants to know about previous medication and allergies. To the scanner, that medication history is irrelevant but they would like to know about scars and previous operations you might have had so they know how to interpret bits that come up on the scan. Etc.

Grass roots have gotten us where we are today: a very de-centralised mish-mash of particular IT services for particular local needs. Your story illustrates the problems, not the solution.

observa
observa
10 years ago

Patient: What’s the diagnosis Doc?
Doc: Well your symptoms are somewhat cloudy at present so perhaps we’d better wait for the NBN for a more definitive answer.

JMB
JMB
10 years ago

Paul and Nicholas:

Surely the biggest problem is the fact that the patient does not own the in formation. The GP’s notes to file, the history of conferences between treating specialists before and post-op, even down to basic blood pressure records taken by the GP/pharmacist/nurse.

Most of the stuff which ties together the raw data isn’t the patient’s property and, when it comes time for the GP’s practice to change hands, one of the determinants of the sale price will be the number of patients’ files which will change hands.

The question quickly boils down to what is in the interests of those who either own or control the records. They are the ones with the most skin in this game. Anything that tends to devalue the role of a GP (or pharmacist or … anybody’s business or profession) will be resisted fiercely by those same actors.

It won’t happen in my lifetime, unless central government demands that it happen, which they will not. Nicholas says that it cannot happen top down. Paul and I say (for different reasons) that it won’t happen bottom up.

So, it won’t happen.

David Walker
David Walker
10 years ago

My very limited experience is that this is one of the hardest problems around. JM and Paul F are right.

I’ve talked to Microsoft people who found it deeply frustrating. One problem was that there was no obvious place to begin. Last I heard – a couple of years ago now – Microsoft was aiming at the pre-kinder demographic because at least the parents are motivated consumers of health information. Nothing much seems to have come of that.

Like Nick, I start by seeing this as a protocol problem, like the retrieval of online information pre-Web. The right answer to the question “How do we enable electronic health records” is: “Whatever leads people to adopt a common protocol”.

So here are a few directions towards adoption of a protocol. But watch out: my general sense is that most of my thinking just mirrors that of other people – who tried and failed. The fundamental problem here is that we don’t understand what it takes to get protocols adopted in this field.

The first direction is to keep the protocol simple, extensible and maybe human-readable, the way HTML was simple and human-readable. There are already various flavours of Health XML – a markup language for health records – that might fit the bill. They may be too complicated. Nick is right: the solution will be start with something very bare-bones and then let it build.

The second direction is probably to keep the protocol from being the property of one company or government. Even Google suffered from privacy worries over Google Health. Microsoft’s HealthVault is probably incapable of succeeding simply because too many people don’t trust Microsoft. Any government will suffer from political pressures to address local problems in ways which will stop the protocol from working internationally – and a lot of people don’t trust governments with their health information anyway.

The third direction is to ensure an implementation of the protocol will run on fairly simple software. You might create a Dropbox-style system that health services can send your records to, that implements some form of public-key cryptography, that allows fine-grained control over which key-holder sees which record, and that will run on any modern server running Linux or IIS. This is not too far away from the solution suggested by Ed’s Mum.

The fourth direction is to start with a system that will appeal to technology-literate early adopters. If they adopt it, the rest of the audience may follow.

Australia is following a different route, as are many other countries. The Australian government has been trying for many years under Coalition and then Labor to introduce its own system under the NEHTA banner, implementing standards such as Health Level 7. So far this appears to have been a remarkable and expensive failure, but I have not followed it recently. Or perhaps it will come good soon.

Observa is right if he is suggesting that this is a much more important problem for online health than any of the problems supposedly being addressed by the NBN. And it’s interesting to note that speed of data transfer is a low priority here. If I have a 2GB medical scan, in most scenarios a 1MB/second upload speed would be just fine for moving it around. Moving data is a solved problem; getting people to use it well is hard.