Tony,
This is very impressive piece of work. Every since I first came across openEHR I have intuitively felt that it is closer to the ‘solution’ than more static attempts at standardization. So why is progress so slow? I’ve appplied some lateral thinking to this, and come up with what many people on this list may (at best) think contrarian - but at the risk of being flamed…
The Case for NPfIT 2.0 www.nationalhealthexecutive.com page 52-53.
(I’ll go get my hard hat now…)
Best wishes,
Derek.
Tony,
I agree thanks for the work and for sharing.
W. Ed Hammond, Ph.D.
Director, Duke Center for Health Informatics
Derek Meyer
<dmeyer@sgul.ac.u
> To
Sent by: For openEHR technical discussions
openehr-technical <openehr-technical@openehr.org>
-bounces@openehr. cc
org
Subject
Re: Articles on Healthcare,
10/22/2010 12:16 Complexity, Change, Process, IT
PM and the role of openEHR etc
Please respond to
For openEHR
technical
discussions
<openehr-technica
l@openehr.org>
Tony,
This is very impressive piece of work. Every since I first came across
openEHR I have intuitively felt that it is closer to the 'solution' than
more static attempts at standardization. So why is progress so slow? I've
appplied some lateral thinking to this, and come up with what many people
on this list may (at best) think contrarian - but at the risk of being
flamed....
The Case for NPfIT 2.0 www.nationalhealthexecutive.com page 52-53.
(I'll go get my hard hat now...)
Best wishes,
Derek.
All I can say Derek; is that if you think my past medical, mental and
social history older than six months is useless information. Much less
my familial history of a few generations.
I am very happy that you are not my physician.
Maybe if you had all of that information in a meaningful semantically
connected network. You could practice better preventive healthcare as
opposed to band-aid, reactive medicine??? 
Cheers,
Tim
Hi Derek,
it is very simple. Not being an official standard has been a real problem for government agencies, obsessed with official standards.
Tim,
I don’t claim that all old information is useless.
My hypothesis is that clinical care generates vast amounts of information, and very little of this vast amount is useful.
(This is an empirical hypothesis, and so could be measured, although I don’t know of a study that has. Perhaps a study that
a) converts real patient records into facts, and the counts the number of facts,
b) requires patients to be seen without a written health record and a treatment plan formulated,
c) reviews the treatment plans in the light of the written record, and
d) counts facts which result in changes to the treatment plan,
e) calculates the ratio of facts that were useful in altering the treatment plan compared with the total number of facts.)
My hunch is that there are gold nuggets in historical records, but we have to capture and store too many pebbles to get the nuggets we need. If there was zero cost to capture and storage this wouldn’t matter, but unfortunately this is not the case with current technology.
This is an economic problem, and the solution is to look for economic benefits at the other side of the time spectrum. If information could be sent to the person who needs it quickly, this time saving could justify the cost of capturing and structuring the information. Once data are structured and captured, it becomes cost effective to do a large number of other things with these data.
This is not an argument against openEHR - just another way of using openEHR.
Best,
Derek.
I think that the ‘pebbles & nuggets’ characterisation is probably right, although I don’t think anyone knows what the balance is, i.e. at what point it ceases to be worthwhile to trawl back in time. The trouble is you get patients like a 12 yo child with a history of chronic tonsilitis that is only visible by looking at say 10 years of data. Or try the other end of the spectrum - notes by GPs over some years may turn out to be indicative of alzheimers, but only when a diagnostic guideline is applied to say 5 or even 10 years of data. So how far is far enough?
I think that what will be needed in the future is a way of filtering out the useless pebbles on the way so to speak. Perhaps when data were archived onto slower media. I wonder if anyone has seen research to indicate how far back data might be useful based on specific morbidities?
I think that the 'pebbles & nuggets' characterisation is probably
right, although I don't think anyone knows what the balance is,
It isn't even easy to (sometimes not even possible) to know
what are the pebbles and what are the nuggets.
In fact, pebbles may turn into nuggets.
I think that what will be needed in the future is a way of filtering
out the useless pebbles on the way so to speak. Perhaps when data
were archived onto slower media. I wonder if anyone has seen research
to indicate how far back data might be useful based on specific
morbidities?
That probably wouldn't be useful because we don't yet know
which morbidities are going to be relevant for a given
not-yet-patient.
Karsten
I don't claim that all old information is useless.
My hypothesis is that clinical care generates vast amounts of information, and
very little of this vast amount is useful.
Make that "... at any one time."
a) converts real patient records into facts, and the counts the number of
facts,
b) requires patients to be seen without a written health record and a treatment
plan formulated,
c) reviews the treatment plans in the light of the written record, and
d) counts facts which result in changes to the treatment plan,
e) calculates the ratio of facts that were useful in altering the treatment
plan compared with the total number of facts.)
Once it was said "If human beings were alike medicine could
become a natural science."
That is why the above plan is doomed to fail.
This is an economic problem,
Health is NOT an economic problem. Care can be, but health
is not.
Karsten
Sorry, I was not clear enough. I meant: if it could be shown that certain patters over time corresponded to certain morbidities, then in new patients (as yet undiagnosed) these patterns could be detected early on.
I see. I can easily agree to that. But I don't think that it
would warrant classifying any EMR entries as pebbles.
Karsten
Dear all,
I have spent some time studying how doctors used an EPR using log data (Determinants of frequency and longevity of hospital encounters` data use). I must say that some of our results were not so expected, namely the difference on the usage of past information according to patient age (reports of children and older are less used much faster).
I am currently leading a research team to repeat the same study on other logs and so I am very interested in collaborations.
Regards
Ricardo Correia
Interesting comment Thomas,
I think " official standards " have nothing to do with obsession, but with governments that have a legal obligation to ascertain some equality on markets, regulations, and ensuring free access and opportunities for all. Maybe I miss a few here, but I am convinced that at least in democratic societies, it is what we as citizens want.
E.g. EHR laws do require official and public accessible standards from official SDO’s with formal balloting and procedures in place in which all parties concerned can participate.
In particular the obsession might be on market dominance.
I am currently working on an ISO standard. One of the member countries was commenting that choosing one particular approach in this standard might favor that approach, hence blocking free trade.
Vriendelijke groet,
William Goossen
directeur Results 4 Care
Tom,
I think it really does depend on the purpose of further use of the data.
clinical : in th e juvenile care record, data on growth and development and vaccinations are stored for 19 year of follow up (in NL). Then archiving is 10 years from that.
clinical: if a diabetes diagnosis is based on particular blood glucose tests, and still there is no cure, it will last a lifetime. I have not seen research, but assume that the formal diagnoses and the glucose test this was based on, are stored a lifetime.
Statistics: if aggregated from records, depending on architecture decisions, you might want to store the data e.g. on population diseases, over many centuries. Of course the stats themselves are kept in the research db, but how long do we want to trace back the source data for quality assurance and analysis of outliers.
So I agree, we need data.
The only real research in this area I know about is the algorythms for eg monitors that average the heartrate and similar measures continuously. Although never searched this further, just assumed vendors do their jobs.
Vriendelijke groet,
William Goossen
directeur Results 4 Care
I take your points, but there is a clear priority for quality, not just equality. In all other industries, the quality of standards is measured initially against public safety and then against criteria of effectiveness and economic qualities. No standard gets through (or if it does, survives long) if it a) endangers the public or b) doesn’t do its purported job properly. In all other industries that i know of, standards are created by a process whose inputs are already developed and productised offerings from companies (or sometimes other entities, e.g. universities). The process is usually one of choosing or it may be one of a compromise agreement. Whatever the detail, the outcome is usually dependable, certainly in modern times. An ‘obsession’ with standards of this kind would be reasonable.
However, this is not what are produced in health informatics. In our domain, the standards are created in committee rooms, and are issued, pretty much untested, with no real proof of public safety, utility, fitness for purpose, maturity or value for money. And yet the governments who run e-health programmes remain attached to these de jure standards, despite their obvious shortcomings. People working for such programmes have trouble engaging with organisations that produce implementation validated outputs, because use of such materials is not sanctioned.
Until this underlying problem in e-health is resolved by a major reform in how standards are actually produced, validated, and maintained, I don’t see much hope for efficient progress in this domain.
In all other industries, the quality of standards is
measured initially against public safety and then
against criteria of effectiveness and economic qualities.
it seems you mean, by market testing. If not, do you have an example?
In all other industries that i know of, standards are
created by a process whose inputs are already developed
and productised offerings from companies
I presume you refer to non-it industries. In IT the picture is rather
more mixed. You certainly aren't describing the omg process, or the
itu process, or the w3c process here.
A truly valid comparison would be with IT standards in other vertical
markets. Insurance always strikes me as applicable. Do you have any
examples from these spaces?
Grahame
In all other industries, the quality of standards is measured initially against public safety and then against criteria of effectiveness and economic qualities.it seems you mean, by market testing. If not, do you have an example?
well yes and no. Products produced by big companies of course have to undergo all kinds of testing to do with safety. With respect to fitness for purpose, the market will certainly sort a lot out. But to get to market, you have to have completely implemented and productised the offering - which means going way past the paper stage. By the time standards agencies see these things, they are guaranteed to ‘work’, the only question is to do with what they interoperate with.
In all other industries that i know of, standards are created by a process whose inputs are already developed and productised offerings from companiesI presume you refer to non-it industries. In IT the picture is rather more mixed. You certainly aren't describing the omg process, or the itu process, or the w3c process here.
IT in recent decades has become quite poor, no doubt about it. Older standards (e.g. older network standards) tended to have hardware implications, and they simply could not be issued without having being implemented somewhere. In more recent times, W3C does at least manage some implementations of what it issues, but is mainly helped by major tech companies implementing the standards. Nevertheless, standards like XML Schema are still horrible, very weak formal underpinning, and hardly fit for purpose (being a document-based idea trying to satisfy data representation requirements). See http://en.wikipedia.org/wiki/XML_Schema_Language_Comparison .
OMG has better process than any SDO in e-health, but the output is not always that inspiring. UML 2 is awful (try reading the ‘infrastructure’ and ‘superstructure’ specs - you really have to wonder what drugs they were taking), as is XMI. Which is why the Eclipse Modelling Framework (EMF) sprung up in the modelling space - to provide a usable alternative to XMI.
A truly valid comparison would be with IT standards in other vertical markets. Insurance always strikes me as applicable. Do you have any examples from these spaces?
I know a bit about investment, and there is to be sure, less to standardise. The interesting comparisons I think are in construction, mobile telephony, automotive, telecomms, etc. Standards just don’t get issued as paper with no products behind them in these industries.
Well, your specific comments certainly don’t back your general statement up. Looking at the question of the other industries, what specific standard would you point to as an example we should follow, and how was it developed?
Grahame
Well, your specific comments certainly don’t back your general statement up. Looking at the question of the other industries, what specific standard would you point to as an example we should follow, and how was it developed?
None of the standards used in these areas were developed in a committee room with a random assortment of people who turned up a few times a year. Instead, companies (e.g. Ericsson, Morotola, Toshiba, Philips, BMW, etc) created products and brought them to market, and then brought the relevant interoperability specifications to standards forums.
E-health should follow the lead of e.g. the telecoms and computer components industries and standardise on things that actually have been shown to work. As I said earlier, it doesn’t just have to be companies that make things that work. Linux, Apache and the IETF standards came from different places. But in all of these situations, the relevant standards were first validated by implementation, deployment before being proposed as a standard. What is happening in e-health is just bizarre. And the results show it.
And none of your examples are vertical industry IT standards.
Mark Bezzina for Stds Australia pointed out to me that IT
vertical standards are a totally different thing to every other
kind of standard.
You’re trying to portray Health IT as some kind of
bizarre exemption, in that things are totally done
in a weird way. But I don’t think it’s an exemption: I
think most IT verticals have the same problem, which
is that standards are being used as a stalking horse
for research.
Grahame
And none of your examples are vertical industry IT standards.
Mark Bezzina for Stds Australia pointed out to me that IT
vertical standards are a totally different thing to every other
kind of standard.
Telecoms, to take one example, consists of many layers of protocols and technologies, most standardised, which form a very impressive stack. It’s all IT, it just isn’t seen as vertical any more because it is such an assumed part of our technological infrastructure.
In terms of basic development process, I don’t see IT ‘vertical’ standards as any different from any other standard. I just can’t think of any kind of standard at all that should be developed in committees of randomly self-selected participants, many with no design or other professional experience. Let’s be clear - all such standards are engineering specifications of one kind or another, and argumentation by committee simply is not a recognised or valid development paradigm for elaborating any kind of technical artefact. It could be used to discuss one, talk about its requirements, but it won’t be able to build the thing in question. And yet this is what happens in health informatics.
In finance, the standardisation that occurs is mostly by industry agreements, +/- government involvement, to set e.g. agreed classification of industry sectors, security types, and so on. The finally agreed schemes come from existing schemes used by companies in the industry, not from committee discussions.
Much standardisation in the military sector comes from defence forces institutions and supplier companies.
You’re trying to portray Health IT as some kind of
bizarre exemption, in that things are totally done
in a weird way. But I don’t think it’s an exemption: I
think most IT verticals have the same problem, which
is that standards are being used as a stalking horse
for research.
I would certainly agree with this last statement for e-health - and it is a terrible way to do research. I have not encountered it in any other IT area, though.