Greetings,
And welcome to day two and can I give a big thankyou to the schedulers who gave me Kate Lundy as the warm up act. It reminds me of that famous 1967 tour when Jimmy Hendrix opened as support act for the Monkees!
Think of me then as Mike Nesmith, he was the one who had a little bit of talent and enough integrity to leave the band and never come back for any of the sadly numerous reunion tours.
My name is Paul Storey and firstly a disclaimer.
My day job is to manage the listing of medicines on the Pharmaceutical Benefits Scheme. In this role I manage a team of twelve and together we publish all the material to do with the PBS including all the Legislative Instruments that give legal effect to the subsidy arrangements. In addition we publish
www.pbs.gov.au a web site that takes a million hits a month, distribute data to over 30 software companies and while developing new things to do with the data I also handle the day to day liaison with over a hundred pharmaceutical companies world wide, a variety of consumer organisations and most demanding of all with other Government departments.
I am here today on my own time and at my own expense presenting material that was written on also my own time and on my own computer. The views I express are mine alone and should not be taken as indicating any policy or opinion of my Department my Minister or the Government more broadly. No public servant has been harmed in the making of this presentation and I’d like to keep it that way.
I occasionally write under the nom de geurre Perry White on my blog the Toy Town Times and I have a personal website
paulestorey.com which I am building to display my sanity preservation device, photography.
I believe in public service and in serving the public good despite having been in previous lives, a puppeteer, a music promoter and enslaved to a venture capital company and forced to live in Toronto for a couple of years.
Most importantly I am completely the wrong person to spruik the glories of the new internet technology and the social media revolution as generally speaking I am not too fond of humans even at the best of times. I do have a Facebook account, but it has the maximum security settings so if you are not a friend of mine in the real world you are unlikely to become one in that world. I think second life is one life more than I can handle and I will close my My-space account as soon as cute young Russian girls stop sending me messages telling me how they found my profile by accident and I look like the type of “guy” who would make them happy for the rest of my life, which by the look of them is likely to be a considerably shorter time than theirs.
I do buy some stuff on EBay, I buy some tunes on ITunes, and get to read lots of interesting material referred on by my fellow tweeters but basically what I am doing in a not too sophisticated way is buying something from a store, listening to the radio and visiting the library.
Developing ways of doing something faster or with greater efficiency is good of course in so many ways but it is not the same thing as doing it smarter. It was said that if the Government had been put in charge of finding a cure for tuberculosis (TB) we would have got the biggest shiniest iron lung you ever saw but not a vaccine.
Let me illustrate, The Shedule of Pharmaceutical Benefits is available in many forms, one of which is a printed book. In November it will be 980 pages. Prior to December 2006 it was a printed publication only and came out three times a year. The "yellow book" as it was known needed to contain information on what drugs were approved for subsidy and the conditions that attached, with me so far, good.
To achieve this a database was designed with one function, to produce the text files necessary to provide the printer with camera ready copy.
Naturally when we needed to move the "book" online to achieve monthly updates the brief to the developers was not to gather the available data into a manageable environment, it was simply to build a website that carried the information found in the book but to do it on a screen. Planning was well advanced by the time I got my hands on the project and at that time an enormous amount of handwringing not to mention development time had gone into constructing a severely limited web presence.
Most of you will be familiar with the Matrix movies, in the first movie there is a scene in which the Keanu Reeves character starts to see binary strings rather than objects. In the Department of Health I am that guy. It is all about the data because it is the only commodity we deal in. It's the data stupid.
Often we (Government) imagine the best way to improve a process is by adding more process, like putting another mirror in a car to get better rear view. Although we are the holders of vast amounts of data I would suggest that even if we were pretty damn good at packaging that data which clearly we are not, and even if we really could and wanted to do interesting things with it that does not mean that we have sole right to the data or should be ones who determine how it can be used or by whom except in some pretty obvious circumstances. This is now thankfully more or less the position of Government.
Can I have a show of hands please if you have heard of Tim Berners-Lee, okay that is a bit of a no-brainer, but now can you please indicate if you know who Ted Codd was?
Thank you, there will be no more deliberate audience participation I promise.
To my mind there have been just two important advances in computing since the 16
th century if you don’t count electricity. The first is the web and the second is the invention of relational databases.
The history of the relational database began with Codd's 1970 paper, A Relational Model of Data for Large Shared Data Banks. This theory established that data should be independent of any hardware or storage system, and provided for automatic navigation between the data elements. In practice, this meant that data should be stored in tables and that relationships would exist between the different data sets, or tables.Initially IBM, where he worked when developing this theory, refused to implement the relational model in order to preserve revenue from their existing information system; one that they had developed initially for NASA as part of the Apollo space program and imaginatively called, information management system database one IMS/DB1.
IBM's relational database system IMS/DB2 was not introduced until twelve years later in 1982 and then only because of market pressure. You should read about the history of database development in your own time but with hindsight the potential of relational tables when combined with the web is mind numbingly obvious.
I called this presentation "The supermarket in the clouds" because in a very real sense my role as a publisher of Government data can be most illustratively described, at least in part, as being a mere shelf stacker in what should become a comprehensive data store,
data.gov.au or at some time in the future data.gov.global
Essentially I produce ingredients, I have plenty of ideas about how the ingredients I produce can be put together to make something of greater interest and appeal, but I am sufficiently wise enough to know that others may have different and better ways of taking my small contribution to the data store, matching it with other ingredients and coming up with something quite surprising.
It is a disturbing habit of mine to use analogies such as this to bring quite complex notions into focus, primarily for my own benefit (of course) and hopefully by using a few more I will shed some light on how in my small area of Government we are moving towards a more useful data model using the power that is slowly becoming available to us.
I say slowly advisedly, because my work computer is a 2001 IBM Thinkpad running Microsoft 2003 with a gut wrenching 500 meg of RAM until I boosted it to 2 gig. The biggest file I can email has to be less than 18 meg and I often don’t find out if it has been sent or not until the next day. Our email system returns out of office replies from my colleagues bundled together at 4.30 every day sometimes up to nine hours after I sent the email, the office building that I work in is less than 12 months old and I cannot get mobile reception unless I stand next to a window and the internal WIFI which was fully installed over a year ago has not been activated. I could go on about our IT services which are provided by the same people who gave Ted Cobb such a hard time, I won't but I am sure Ted feels my pain!
In reality I work in a small area of E-health that will hopefully contribute to developing opportunities for web based and mobile applications, research opportunities on drug utilisation and facilitate conversations on evidence based health by engaging the community in the discussion on why and which medicines are subsidised. In this context I am just starting to get ready to start using the innovations that we call web3.0 and the powerful communications channels available through social media.
The problem for me is that I am totally dissatisfied with any descriptor I have been presented with for either E-health or Web 3.0. So in the absence of consensus I will step bravely forward to fill the void with my own inadequate understanding.
Web 3.0 is commonly called the internet of things; to acheive an internet of things we need to develop IPv6 and quickly, because we are fast running out of the 4.2 billion addresses available under IPv4. 4.2 billion seemed like a big number back in 1981, but who knew? I can live with this definition of an internet of things, even if I don’t fully understand what it means, and I have already given some of my thoughts on social media when people use it, but the real evolutionary leap in my mind, comes when we put the web and relational tables (which use a common semantic) together and get a social media for things or if you prefer a social media for machines.
John Dillinger (who was for a time the FBI's most wanted man) was once asked "why do you rob banks"? His answer was simple and concise; he said "that's where they keep the money!"
My point is that often we make the question more complex than the answer. In my world, the Pharmaceutical Benefits Scheme, we spend in the region of $9 billion dollars a year subsidising medicines. No one buys drugs without hoping for some benefit and clearly no government is going to shell out $9bill without getting some kind of return, in short the benefit to the community has to be greater than the cost, or deemed "worthy" of the cost.
It may surprise you to know that the PBS is not a drug subsidy scheme at all. We do of course subsidise drugs, but solely to make them affordable to everyone so as to deliver a health outcome.
We measure these health outcomes in QALYs, that is quality adjusted life years, and ICERs or incremental cost effectiveness ratios. Put simply, we put a dollar value on how much longer you will live and with what quality of life if given a specific drug and we measure whether that cost is more or less than getting the same health benefit by other means. If the Pharmaceutical Benefits Advisory Committee, an independant expert committee determines that a drug is both affordable and effective, if it is cheaper in delivering the same result which is delivered by other means Govt will generally subsidise it. And in its deliberations the Committee should and does take into account the impact of the treatment on patients. Equity of access is fundamental to the PBS, but factors beyond impact on individual patients are important considerations.
So by virtue of this process I have two things;
- a detailed comparative analysis of the cost effectiveness of a medicine compared with other means (if they exist) of delivering the same benefit, or with a placebo if it does not, and secondly
- a set of conditions under which the Govt will pay a substantial amount of money to treat a particular and very precise set of medical circumstances.
So what does any of this have to do with web 3.0 and social media?
To say simply “I don't know” is not really correct, “not sure” would be closer, to say “well I have a few ideas but others will be able to do things with that information I cannot imagine” is probably the closest I can come to the truth.
My colleagues around the world are working on global harmonisation of Health Technology Assessment, that is achieving a standard for the evaluation of the cost effectiveness of treatments. This will mean that some day the Scottish assessment will work to the same set of rules as the Polish, Canadian, and the rest of the world and possibly even the Americans.
The second means that I can ideally collect information on which drugs have been prescribed in what way to treat which disease. This is only a short step away from doing some fairly interesting analysis as to how cost effective they are in the real world. My little chunks of data go into the "supermarket of Gov data" and then the fun can really begin.
Shakespeare wrote;
Shall I compare thee to a Summer's day?
Thou art more lovely and more temperate:
Rough winds do shake the darling buds of May,
And Summer's lease hath all too short a date:
Apart from the "thees" and the "haths" he uses words that are commonly used by all of us but by cunningly rearranging the order he comes up with combinations that make lovers weep and thousands of school children yawn and look out the window.
My language is
SNOMED CT, the systematised nomenclature of medical and clinical terms, it is the Australian Government standard; it is a concept based numerical taxonomy that allows me to map Myocardial infarction to heart attack to ataque del corazón which is Spanish for the same thing.
With SNOMED CT I can use a numerical hierarchy to describe a second degree deep tissue burn between you second and third toe on you left food treated with antibiotics available in blister packs of thirty prior to amputation after the onset of gangrene…. You get my point.
Once medical, medicine and treatment terminology is concept based and its descriptive terms are fully mapped to the underlying numerical taxonomy AND universally applied we can then use those terms to give the level of precision needed to meaningfully interrogate health data. By adopting a universal nomenclature, (SNOMED CT currently has over 2 million mapped terms) we can have phase one two and three trial data, doctors notes, patient records (including personal E-health records), pharmacy dispensing systems, hospital patient management systems etc linked to each other in a way anticipated by Cobb in 1970 and then all the machines holding this information will be able to go to the machine equivalent of Facebook where they can talk about Paris Hilton and Ashton Kulcha whoever the hell they are.
This notion of linking data sets will unleash an avalanche of data, but to what end? Let’s look at one simple example; Rofecoxib, is a non steroidal anti inflammatory pain killer is one of the COX 2 inhibitor group, it works great in reducing arthritic pain but can elevate the risk of heart attack. Rofecoxib was marketed by Merck as Vioxx, and was on the market for 5 years from 1999 to 2004. Worldwide over 80 million people were prescribed Vioxx at some time. It was thoroughly tested and evaluated, given market approval by the FDA and other regulators and was widely prescribed because it worked, except for the heart attack thing but who knew? In the year before withdrawal, Merck had sales revenue of US$2.5 billion from Vioxx
FDA analysts estimated that in the five years the drug was on the market Vioxx caused between 88,000 and 139,000 heart attacks in the US, 30 to 40 percent of which were probably fatal. Thirty to sixty thousand dead bodies is what I would call a negative performance indicator. The drug was withdran world wide in September 2004.
On November 5 2004 the medical journal
The Lancet published a meta-analysis of the available studies on the safety of rofecoxib
(Jüni et al., 2004). The authors concluded that, owing to the known cardiovascular risk, rofecoxib should have been withdrawn several years earlier.
The Lancet published an
editorial which condemned both Merck and the FDA for the continued availability of the drug from the year 2000 when the cardiovascular impact data was becoming increasingly available, until its withdrawal in 2004.
Merck’s response was to issue a rebuttal of the Jüni
et al. meta-analysis that noted that Jüni omitted several studies that showed no increased cardiovascular risk. (Merck & Co., 2004).
The point is that eventually the data established beyond doubt the correlation. My question is this; with a common nomenclature and the means to link and interrogate the various dispersed data sets would it have become obvious sooner? I believe so. But more importantly we should be asking about other evidence that could be available if we had integrated data sets and decent analytical software? Patient has heart attack and dies is one key performance indicator but dramatically misses the point.
Another
example; Dr. Rachel Freedman is a medical oncologist at Dana Farber Cancer Institute in Boston, and an instructor in medicine at Harvard Medical School.
Examining the delivery of breast cancer treatment this year 2010, and adjusting for variables such as age, income, education and insurance coverage, Freedman found that black women were 9 percent less likely than white women to get mastectomy, breast-conserving surgery or other surgical treatments, 10 percent less likely to get hormonal therapy and 13 percent less likely to get chemotherapy.
She now recommends that women diagnosed with breast cancer write a list of questions before they go in for a doctor visit, take a friend or family member with them to help them understand options, and consider taking a tape recorder so information can be replayed later.
How long I wonder, before a video of the visit to the doctor becomes part of the patient record together with a video of your improvement or decline during the treatment cycle made during self monitoring and assesed by your medical provider?
Linking patient information in populations big enough to yield statistically meaningful information could lead us into a whole new set of opportunities, not only in the reporting and interpreting adverse drug reactions or differential care models but in giving unexpected beneficial correlations like finding out that half an asprin tablet reduces the risk of heart attack.
This is where web3.0, a social media of machines and decent interrogative software will deliver things we cannot contemplate.
Now if you will, permit me a bit of future gazing. I believe that probably within my lifetime individual genome mapping will be part of the patient record from birth doing away with genetic testing for individual targeted therapies.
Intel is currently developing software that measures cognitive decay by measuring how long it takes you to recognise the caller on the other end of the phone call. It measures response times in fractions of milliseconds and looks to be a predictor of the onset of dementia about five years earlier than existing diagnostic tools.
If we could link patient lifestyle data, medical record and personal genetic map to dementia onset, or any of a number of other conditions, a comparative analysis of the data could put us one step closer to an effective treatment or better still prevention.
The evolution of E-health will bring numerous challenges not the least will be the ethical questions, primarily privacy and the right to know. How we meet those challenges will be a test for us as a community.
My current area of interest is in optimising the effectiveness of drugs, this is leading into some pretty interesting areas of biomedical engineering particularly in implantable narrowcasting patient telemetry and dose delivery devices. Try saying that after a couple of vodkas! These are nano machines that you will all one day wear on the inside and which will give a continuous readout of a whole range of body system information. We can interpret those data and then regulate the flow of medication where and when you need it, for example automatic release of adrenalin with the onset of a severe allergic reaction or insulin at a certain blood sugar threshold. We know that variations in blood sugar levels are also an indicator of possible infarction but monitoring is haphazard at best. It is also possible to read and record such things as blood oxygen content and the rate at which you are accelerating. This is important because if your blood oxygen level drops as you are accelerating towards the floor at 3 metres per second per second it is a pretty clear indication that we better get a crash cart round to your place pretty damn quick.
In the context of a residential aged care facility the choice is stark; it is either immediate notification and rapid response or finding out next time you don’t show up for volley ball or bingo. For stroke victims that can be the difference between a short hospital stay and full recovery or a longer hospital stay and probable death. Believe me I am at an age when these thing start to acquire a certain clarity if not urgency.
Early intervention saves lives, saves quality of life and for Governments saves billions of dollars. Determining the type of intervention will be evidence based, it is the data stupid, I can only provide a small piece of that puzzle. We are at the start of the journey and in the immortal words of Bette Davis "Fasten your seatbelts—it's gonna be a bumpy night!".
Archimedes said “give me a lever and a place to stand and I will move the earth” I say give me the data and the ability to interrogate it and I will give you a better and longer life.
A lifetime ago when I was promoting rock and roll I would talk to musicians about performance. Some like the Mentals Oils Chisel etc would work so hard on stage that the audience would leave the gig emotionally spent. Others with names that are long forgotten would believe that the transformative power of their songs was sufficient and they merely had to play them. They didn’t understand that anytime you are on stage it is a performance, the only measure is the degree.
Today we in Government are engaged in social media and dare I say it, developing the potential of web3.0. What is yet to be judged is how well we are engaged and how good we are at it. Our challenge is not just to make the data usefully available to developers but also to work at improving health literacy in all areas and delivering that knowledge not just through our own activities in social media but through engagement wherever the opportunity arises to encourage people to become actively involved in their own health and in the health of those around them.
Surely there is some social obligation on us as a society not just to collect our medicine but also to enable our individual experiences to be aggregated to prevent the next Vioxx and to deliver the next benefit.
I will be happy if we simply start to imagine the possibilities. As Wallace D Wattles said “nothing that man can imagine is impossible of realization” but perhaps Shakespeare said it best There are more things in heaven and earth, Horatio, Than are dreamt of in your philosophy.
Oh and this morning my data set got a message from a cute young Russian forecasting model that came across him by accident and she thinks he’s pretty damn cute.
Thank you
