Sunday, January 05, 2003
The single health record conundrum
I'm delighted to see that Phil Windley is continuing to post now that he's moving on (or even moved on already) from his post as CIO in Utah. One of his recent post poses the question "why isn't there a single health record", to help avoid adverse drug reactions (in his case, his mother-in-law is allergic to ibuprofen, something which I imagine is relatively common). He notes that there are 2.1 million adverse reactions annually in the US (I've seen figures for the UK suggesting that our equivalent is in the upper single digit hundred thousands). So, given the solution is "all IT and nothing to do with health", couldn't there be one "medical record could be kept in a single location and accessed and updated via the Internet by each doctor, hospital, emergency room, EMT, and pharmacy you use.". This is one of the holy grails of patient care. Not only would it reduce things like adverse drug reactions but it would mean that doctor's would have access to the latest test from the hospital, the same questions wouldn't be asked over and over, patients wouldn't have to carry x-rays or MRI scans around with them and so on. The productivty savings are enormous once its done. In the UK we have a specification for such a single record out for consultation at the moment - I will be following with great interest how this progresses, as the first consultation ended with a need to do more work. There are two big issues stopping it today (in the UK certainly but I fully suspect that the same is true in the USA): - Rationalisation of identities. Use of a single number is inconsistent, even though everyone has an NHS number. This is getting better today, with all GPs using it and from around March this year it will be mandatory across the board. But the history is not going to be easy to reconcile - and things like Ibuprofen allergies may last have been written down 10 years ago or more. - History is on paper or on many systems. To get to that single record, you have to make an early choice about whether you are going to store it all in one big central place or link up a variety of databases to create a virtual record. Either one is feasible - with the former you have to move all the data to the middle rationalising it in flight; with the latter you have to link a lot of old-style systems, maintain the links and probably put some serious bandwidth in to make sure that images can fly around. But both involve similar problems - you still have to rationalise the identifiers, you have to figure out how to get the data out of where is (and whether to transpose it if its on paper), put bandwidth in to link the systems (or to ensure that doctors can download the data). My vote is for the central database, but with cached copies locally (things like X-rays don't change a lot) so that there is not a huge waste of bandwidth moving things around (there could be a need for 100MB/s to do some of this). Of these, the former is relatively trivial - it's a rule and people have to follow it. Data gets better the longer you wait before doing the second action. When I was in San Francisco in December I met with a great company, Zmedix, that have a "patient questionnaire" application (called CLEOS) that could create a one-time patient record. Picture the scene of a doctor's surgery with WiFi and tablets installed - the patient goes through the questionnaire which covers pretty much everything that has gone on with their medical history and all recent things with a structured set of questions (potentially collecting up to 35,000 datapoints, for someone who had a lot of history!). A report is then spit out which the doctor can use to evaluate what to do next (and doctor's need some help here - they don't have the time to spend with every patient, the situation is often complicated by some symptons masking others and a patient desire not to always tell the truth) ... and a detailed electronic record is created which could be uploaded and stored centrally. All you have to then is map images, correspondence and so on to that central record. It feels to me that the solution is in sight, but it's going to take some creative steps to get there along with some dramatic changes in working practices. Whether it is CLEOS or some other equivalent, doesn't matter. The benefits are so obvious, why would you not do it.
Posted by Alan at Sunday, January 05, 2003