Category Archives: it’s getting better & better

An Open Source Advocate’s View of AMIA 2003

I had the opportunity to attend the AMIA Annual Symposium last week.
Ignacio has already reported on some of the events but I would like
to point out some of the other sessions and the general attitude about FOSS at AMIA.

This is my 4th visit to the Fall AMIA Congress as an open source advocate
and it is encouraging to now meet people and have them not give me a
dumb-founded look when I mention that I promote open source use in
healthcare informatics. So this is progress!

Open source in general has a real presence at this year’s conference. First of all there is an entire page of the program highlighting the various open source related presentations. The mysterious AMIA open source task force is still in existence and has apparently been working in stealth mode according to Martha Adams of Duke University. According to Dr. Adams they have been busy determining what manner AMIA could properly get involved in open source.

They are now prepared to move to the next level and become more open and
collaborative with the wider open source community. While I could speculate on the direction(s) this might take I have been assured that the panel discussion on Wednesday (S89) will bring to light the past work and future direction of the AMIA open source effort.
I am waiting in anticipation.

My brief tour of the exhibit hall on Sunday revealed the standard array of vendors normally present at an AMIA conference. Notably missing are the huge booth displays from EPIC and others. Though some of the other large systems vendors are here, the booths are much more subdued than at past events.

The Keynote Presentation by Julie Gerberding, MD, MPH; Director of the US
CDC was quite interesting. She described the work and importance of
information systems during several of the recent threatening health related crisis. Of really geeky interests is tying the data collection systems to the GIS systems for mapping the outbreak links. I would assume that several of the systems use a variety of open source components I do not yet have any hard evidence of that. If you know of any please add comments.
Of special note here as well was the ability to exchange data worldwide via secure network. This was especially helpful during the SARS crisis. Though she did note that there is much room for improvement on all these fronts.

I had an opportunity to meet with Dr. Peter Murray from CHIRAD (UK). He
held a tutorial on open source. His focus was to introduce the definitions and deliver a basic understanding of the concepts to those just becoming interested in the benefits of using open source applications. Peter is also involved in the startup of the IMIA open source working group.

As a member of the Clinical Information Working Group I attended the
business meeting Sunday evening. The group was surprisingly small
especially since it has now been merged with the Quality Improvement Working Group. While it did not start out with much open source awareness at all I believe I was successful in planting the seed that we should look at the possibilities that open source can deliver. The general attitude and points of discussion are certainly based on the software as a commodity paradigm. The project discussion centered on the difficulty involved in selecting an existing boxed CIS and justifying the cost. This meeting made me even more aware that any thoughts of open source being commonly discussed hope that with the proper presentation of open source development this will change but I am resolved that it will
be a long time coming.

The Panel S16 on Monday morning was interesting in the fact that this panel
has been held six times. The very same panel……slightly modified I
presume to cover the latest in patient record deployment development at each of the
four sites. These deployments are taking years and years to roll-out.
This is an interesting panel because it consists of two commercial systems
and two in house developed systems. In all four cases the healthcare systems
are very much a closed loop system so the administration can provide the IT
department with the ability to mandate what will be and who will use it. In
the case of Beth Israel it was specifically pointed out that system use was
not mandated but that the users choose to use it because it is efficient.
This is one of the in house developed systems. There seemed to be no
difference in quality and features of the systems based on the source of
development.

The Epic system had been implemented in such a manner that
there were “Islands of Information” among the various sites/clinics. A
large part of the project there was to bring those islands together into a
Results Reporting facility for the clinicians. The project estimate from
1993 was $30,000,000. I was unable to discover what the real cost to date
is but 10 years later I would guess it is slightly more.

Aurora Healthcare has a (mostly) Cerner system. In addition to the Cerner
suite they use AllScripts for electronic prescription writing.
I will attempt to find out if the
prescriptions are also electronically recorded in the patient record. Using
this centralized system across their organization gives them 16 points of
failure between the clinicians and the database. I think this is and
obvious argument for local management of at least primary care records.

The final system was presented by Dr. Overhage from the Regenstrief Institute
of Healthcare. This in house developed system has actually been in
development since the 1970’s. This system stands out because it has
withstood many technological changes and continued to grow to meet the
user’s needs. While every organization cannot develop their own system a
group of organizations with a common goal can and should be producing an
open system. When one organization demonstrates they can do this in house
then certainly a group should be able to accomplish the same goal.

An interesting couple of comments were that the Beth Israel site cannot
determine who is using the system. I would really like to get this
explained further from Dr. Sands. Also, in all four implementations there
are no data quality audits being performed. The one (kind of) exception to
this is at Aurora Healthcare they have a manual point during the coding
process where two people look at the data.

I attended this set of paper presentations on CPOE deployments primarily for the last one. But the first three proved to be interesting as well. These papers all looked at the data quality and effectiveness of CPOE in a variety of implementations. If you are involved in a CPOE deployment I suggest you read these papers as background.

The final paper in these set was presented by Dr. John Windle from the
University of Nebraska Medical Center. The title; ‘Web-based Physician
Order Entry: an Open Source Solution with Broad Physician Involvement’
certainly caught my eye. This application uses Zope and MySQL running on a
Dell server pre-installed with RH 7.3. This system interfaces with, but is
not tied to IDX. The code has not yet been released but I believe we can
expect it after they move primarily from a DTML & External Method based
application to a Python Product. I found their user interface to be very
clean and functional. There was a workshop on Tuesday morning for this application.
The room was overflowing and there was a great deal of interest from those I spoke with
in using mature open source tools to develop healthcare applications.

No doubt one of the most widely anticipated panels was the AAFP presentation
on their open source project. Dr. Kibbe was the first speaker and he
basically regurgitated the information that has been available in the press
releases regarding the history and failure of their open source effort.
BTW: Failure of the open source effort are his words.
My perspective is that they never had an open source effort. They had a
plan to take a commercial product and release the code as open source though
it is unclear of the details here as one of the slides had mention of
database licenses as well. This gives the indication that this code was
still tied to a commercial database. Either way, Dr. Kibbe would not
discuss the reason(s) he could not get the buy-in required from other
organizations. I speculate that it is simply because Medplexus was an
already chosen starting point and this was never a true open source plan.

Dr. Kibbe went on to speculate that open source may not be the solution
though he said he really believes it can be. I have no clue what that
really means. Then he presented some of the ideas they are working on with
a variety of commercial vendors on open standards because in his opinion
open standards is what is needed and it doesn’t matter if the code is open
sourced.

Dr. Zuckerman was the next speaker and clearly his mission was to scrub the
“open source” phrase from the entire AAFP project. According to Dr.
Zuckerman the goal of the project all along was to deliver inter-operable
EHR and Practice Management systems based on open standards.

Again we heard much about nothing as far as open standards are concerned.
This talk was as if
it had been pre-recorded from 1984. I enjoyed his statement; “Office based
servers are bad because they require too much maintenance.” clearly his
experiences have been limited to unreliable systems. Or his mission is to
push a certain set of products that centralize data. He clearly was behind
the option of using a thick client connected to a remote database.
My overall impression of this portion was that the AAFP should be embarrassed
to have him represent them in this forum.

Dr. Mike Bainbridge from the UK was the next speaker.
I have no clue what he has to do with the AAFP open source project (neither did he when I asked him later) but as
always he is an interesting speaker and brought us up to date on what is
occurring at the national level in the UK.

Dr. John Zapp had a great presentation and pointed out that we are dealing
with a non-system system in the US as compared to the UK where there is a
national mandate for an EHR. He also went over the recent efforts to define
an EHR and set standards for the functional areas. He spoke a little about
NAPCI which is now a non-profit organization. Again the theme was mostly
about open standards and that this is all we really need to solve the EHR
problems of cost, interoperability and data stewardship.

To address this whole new direction of open standards I just want to ask the
question, “Why after more than 30 years do we not already have patient
record standards?” The answer most often heard is that it is a very complex
problem. This is true, but we have solved complex problems before. The
real reason is because it is not in the best interests of the commercial
vendors to work together in an open and collaborative manner to make it
happen.

As disappointing as this panel was we can all be assured now that there is
not and never was a true open source EHR plan from the AAFP. It seems to
have been an attempt to hijack the phrase to promote the distribution and
use of one particular product. If Medplexus wanted to open source their EHR
today they could do so without the backing or financial incentives of the
AAFP and other organizations.

If Dr. Kibbe changes his mind and decides that an open source solution is a
good thing for the AAFP then I suggest he learn a little more about the
concepts and processes involved in open source community building before
attempting it again.

The final panel I attended was S89. For a panel held on the last morning
of the conference it was very well attended.
The panelists weren’t necessarily participants in the AMIA OS Task Force announced
last year but this panel discussion was really the report from that Task Force.
Of interesting note one of the panelists is from a little OS project you may know
about called Apache. Brian Behlendorf of Collabnet, Inc. did a great job of presenting
the history of how Apache was started and how open source projects in general may function.
I did note however that he focused on projects of a more horizontal nature and where the users are developers.
In the case of healthcare and vertical market software as a whole the model is slightly different but very doable.

In the end this panel session simply did not have enough time to cover everything.
I do believe the end result will be a very proactive attitude within AMIA regarding
open source software in healthcare.

Overall, among the people I chatted with in the hallways, I found there to be much more awareness regarding open source software
especially with Linux and Apache. I am looking forward to MedInfo2004 (supersedes AMIA’s Fall Symposium)
in San Francisco (see http://www.medinfo2004.org).