Well now if you know that Epic is paying KLAS, do tell, and give evidence! Or is this another Oracle…
Monday Morning Update 1/19/09
From MM: "Re: home grown systems. Foul! When I read people call software built by healthcare providers as ‘home grown,’ I can feel the hair on the back of my neck rise and my blood pressure go up. In the 1980s, that may have been an accurate statement, i.e. that providers built ‘home grown’ systems, where ‘home grown’ is used as a derogatory statement that refers to software that was developed by a single person using outdated technology with limited testing and no documentation. Today, there are some leading healthcare organizations designing, building, and deploying commercial grade software. With the move towards ‘Agile’ development, one can no longer equate the number of developers with the quality of software. Another point is that the software developed by a healthcare organization is ‘lighter’ and simpler than vendor software because the requirements are lighter." See below for an example of this from MD Anderson, which hardly fits the ‘home grown’ label.
From Salesgal: "Re: EEOC. [EMR vendor] is under investigation by the EEOC for unfair layoff procedures during the June 2008 layoff. All of the complaints are lodged in [sales manager’s] district. Seems he let go pregnant, ill, and staff who were not his groupies and kept his favorites who were not selling. He was ultimately let go, but several reps with solid experience and sales numbers were lost." I’ve omitted the names since I have no official documents to back up the statement. Sounds juicy, though.
From Rowdy Piper: "Re: layoffs. Orlando Health (formerly Orlando Regional Healthcare System) has ‘reorganized’ positions across the institution by eliminating positions and placing many of those affected employees into other (many times lower paying) jobs. Additionally, they have instituted a system-wide hiring freeze and have started to cut hours for many and even forced time off over the holidays."
From Corndog: "Re: UHIN. Very impressive that Axolotl beat out Medicity right in its own backyard. Intermountain Healthcare has been shrewd in using technology to advance the quality agenda and must have liked what it saw." Maybe, although its big GE partnership doesn’t seem to show infallibility. Still, Axolotl has strung together some pretty good wins recently.
The President-elect is still talking up healthcare IT, this time in a TV interview: ".. in the economic recovery package that we put together, we have a lot of investment in making the health care system more efficient. Those are things that had to be paid for anyway. Just a simple thing like converting from a paper system to electronic medical records for every single person can drastically reduce costs, drastically reduce medical error, make not only health care more affordable, but also improve its quality."
HIMSS Analytics brags (confusingly) that "85 percent of hospitals in the contracting phase of an IT acquisition have signed with a CCHIT Certified vendor since November 2007." There’s no such thing as a certified vendor, only certified products, and even that certification only covers EMRs, not most of the software applications a hospital uses. Does that mean that 85% of hospitals that have bought any IT system in the past year also bought a certified product, or that those who bought a product covered by CCHIT (inpatient or ambulatory EMRs) certification chose the certified one? If they only bought a server (which is an IT acquisition) and no software, how were they counted? Mumbo jumbo aside, it doesn’t really matter — everybody shamed the vendors to lay out the cash to get certified, so it’s not like hospitals have much of a choice other than in office-based EMRs, where they might pick a lesser-known vendor despite dozens of certified ones. It’s not like having certified products available opened the floodgates – they’re buying the same old products that just happen to be certified now. Results, not surprisingly, haven’t changed – just the cost. Products are interoperable, but users aren’t.
And speaking of CCHIT certification, Allscripts Professional earns Ambulatory 08.
UC Irvine Medical Center is proud that its anesthesiologists have stopped falsifying surgical records by filling out forms before the surgery starts as CMS found earlier, now thwarted by its new SIS software the prevents them from doing so.
Former Healthlink VP Ed Kopetsky is named CIO of Lucile Packard Children’s Hospital.
Jonathan Bush and David Brailer were on CNBC. There’s not much new since the TV guys require everything to be simplified so they can understand it, but I bet JB can’t wait until Tuesday when the talking heads will have to drop their "president’s cousin" knee-slappers unless he’s also related to Obama.
Conditions are so bad that Columbia St. Mary’s (WI) intentionally slows down its $417 million new hospital project that’s already almost four years old.
Big layoffs for Wellpoint: the struggling insurer will drop 600 employees and another 900 unfilled positions.
GE Healthcare will lay off in "the low to mid-single digits" as a percentage worldwide.
Hospital layoffs: Tulare District Hospital (CA), no number given; Excela Health (PA), 70; Wellmont Health System and Mountain States Health Alliance, 195; Fox Chase Cancer Center (PA), 80; Montgomery Hospital Medical Center (PA), 17. Brazosport Regional Health System cuts the hours of hourly employees and issues a mandatory 10% pay cut for salaried ones. There are more, too many in fact for me to keep writing about, so suffice it to say that nearly every hospital is freezing discretionary spending, cutting capital budgets, and laying off staff.
West Virginia University Hospitals and its physician group were to have gone live with Wave 2 of their $90 million Epic implementation over the weekend. That’s $18,000 per user or $180K per doctor, just in case you were wondering.
Sen. Chuck Grassley’s investigation turns over another hidden fact: the orthopedics chair at University of Wisconsin-Madison pocketed over $19 million in five years from device maker Medtronic. He claims the money was royalties for patents he holds. So how the heck was he able to keep all that money without the university knowing (that is, unless he was intentionally cutting them out of the deal even though he’s a full-time employee)? He claims he fully disclosed his relationships and never implanted any of the devices in his own patients (so does that mean their care was better or worse?) In fairness to him, the university’s disclosure forms have a top category of "over $20,000," so that’s their fault for not being more specific.
I’m puzzled: why did the National Research Council and then the Senate ask Microsoft (and Intel, for that matter, in NRC’s case) for its opinions on healthcare and technology? Peter Neupert gave the company’s view (warning: PDF) of what the future of healthcare should look like, but everywhere I’ve worked, they just sold expensive, trouble-prone IT plumbing and gave CIOs free trips to Redmond. Of course Microsoft envisions a radically different and technologically future – that’s their only hope for elbowing aside companies like Meditech that have been automating healthcare while Bill Gates and Jerry Seinfeld were still in high school. I’m not saying they shouldn’t have an opinion, only that I don’t get why anyone would give it much value when they’re dabblers at best. At least its rumored layoffs apparently didn’t happen.
Two 19-year-old students develop Refill300, a free e-prescribing and refill site.
AT&T is developing software that will use both WiFi and a low-power technology called Zigbee to send home monitoring information to providers.
Amazon Web Services makes some of its huge public data sets, including genomic and census data, available at no charge for developing cloud applications.
Three UK hospitals are rolling out e-prescribing in what’s called the UK’s biggest eRX project. JAC Computer Services, a subsidiary of Mediware, is providing the application.
M.D. Anderson’s EMR Project
By Lynn H. Vogel, Ph.D., FHIMSS, FCHIME
Vice President and Chief Information Officer
Associate Professor, Bioinformatics and Computational Biology
The University of Texas M. D. Anderson Cancer Center
I asked Lynn Vogel about the project they’ve been working on at M.D. Anderson. Here is his response.
The January issue of Advance will have some comments on what we are doing with regard to ClinicStation and its SOA architecture, with some observations regarding comparisons with commercial products.
The fundamental problem is that every commercial clinical systems product on the market today relies on a single, physical data repository, generally with HL7 interfaces, and is based on relational data base architectural models that are now close to 20 years old. Outside of healthcare, the IT world has embraced Services Architectural models and is now moving into semantic data models as well. But the cost of a major architecture change is simply prohibitive for commercial healthcare IT vendors. For all the talk about interoperability challenges, not being able to incorporate new data models into our clinical systems environments down the road may be a much larger constraint on improving our nation’s health.
The advent of genomic (or personalized) medicine presents very different types of data from what we have been accustomed to historically. For the most part, clinical data has been viewed as a series of discrete data models — you have a particular laboratory value, or a radiology report, or a graphic from an echocardiogram that you send to a repository via an HL7 message — so incorporating new data of this type into the repository via HL7 has not been a problem. But genomic data models tend to be based more on pattern-matching than discrete data, and products that rely on a single physical data model have real problems integrating these new data types. In addition, the vast volume of genomic data that is now being collected and increasingly available for matching can become unmanageable within a single data model and repository structure.
M. D. Anderson is now working on a parallel product, ResearchStation, also built on the SOA framework, that for the first time promises true integration of research data (e.g., biomarker data) with data from clinical practice. We hope to exploit UDDI capabilities as the actual linkage process between data and services from ClinicStation and ResearchStation. All of this, of course, says nothing about the vocabulary and terminology challenges we face in healthcare, for which is there is simply no comparison in any other industry. This is where semantic models promise to make a significant impact — and major technology vendors such as Oracle are already incorporating such capabilities into their product suites — but these are not even on the radar of the major healthcare IT vendors.
A couple of references with more specifics on what we are doing:
Case study from Microsoft (Word document)
Case study from Avanade, our strategic partner in software architecture and development
Wes Rishel from Gartner included our work as a case study in his presentation a couple of weeks ago at the Gartner Healthcare IT Symposium, so I think the word is getting out about the usefulness of SOA and how it can help us to deal with a number of the data challenges in healthcare.
Re: Dr. Vogel’s comments – very clearly articulated. As budgets dry-up and monolithic implementations continue to be challenged, the build strategy will re-emerge. It always does. Larger organizations will turn toward strategic vendor partners to help build key portions of the architecture. Examples would include a Swiss Army Knife clinical data repository, visualization and integration layers, orders and general documentation functions. This core would be surrounded by a set of best of breed applications. This can work well until you encounter massive changes (ex: regulatory changes, paradigm shifts = genomics). A similar argument is made here:
http://www.healthcare-informatics.com/ME2/dir
mod.asp?sid=349DF6BB879446A1886B65F332AC487F&
amp;nm=&type=Blog&mod=View+Topic&
mid=67D6564029914AD3B204AD35D8F5F780&tier
=7&id=1CF51170E5FB4BE1B1F87F135927B254
Where my opinion may diverge is that I do not think it is feasible to define a SOA among a multitude of vendor environments where you do not control the “core.”
I wonder how MD Anderson is dealing with the data discontinuities discussed by Dr. Glaser.
Relating to the Brailer / Bush CNBC interview Anyone have a clue what Jonathan was trying to say? 50% of everything doctors order doesn’t come back within 30 days so they reorder it? Therefore they have a 50% error rate.
I believe what he means is athenahealth takes on the populating of the patient chart as a service for its clients using athenaClinicals. So unlike a traditional software EHR where the vendor may have no idea how the client is using the system, athena is responsible for tracking down labs ordered etc even if the lab is no online, which obviously means athena then needs to build an electronic link to chip away at that drop off rate. Otherwise, the client and the staff would need to either build an interface or keep waiting for the labs to come back via fax etc. From what I know of the product once they build an interface to a lab it benefits all users on the network even if they don’t use that lab. over time the drop off rate goes down. They have been leaders at this on the rev cycle side for years and are applying the same model on the clinical side – quite a bit of heavy lifting but the value of being on their network goes up as they build increased automation – think of your bank building interfaces to all the places you bill electronically rather than you doing it using the bank’s software – imagine how that would be??!!
Think of the rev cycle world, they must still have to deal with payers that still do things on paper etc as part of their service compared to a PMIS vendors that leaves it to the client to deal with the payer or try to connect etc themselves.
I think the CEO wrote about this on HIStalk a while back. I think it is pretty cool they even know, as then they can build process improvement for all clients on the network.
RE: Comment on the feasibility to define a SOA among a multitude of vendor environments. The vendor environment for the last twenty years has been dominated by best of breed legacy systems. The “core” behind most vendor environments can be treated as an exposed service on a SOA platform. If you recall, a(an) SOA is an IT mentality/process to leverage “e-in situ” resources with solutions including the kitchen sink to produce true interoperability.
This might appear to be a naïve observation but it does reflect the simplicity of the HIT environmental conundrum as the sum of the total parts. I call this presentation a lesion by pattern recognition requiring SOA oncology.
I read with great interest that article by Dr Vogel, but looking at the Microsoft white paper pulled back the green curtain. That is to say, I didn’t note anywhere in the article that ClinicStation is in fact a “clinical information retrieval system.” Unless I am mistaken, ClinicStation is a SOAP portal that allows clinicians to view data from existing systems such as IDX and the pay-per-view Stentor.
I will not denigrate the need for data access, but were I Dr Vogel, neither would I denigrate HL7, relational databases, nor interfaces if my solution is “read only.” If I am a physician reviewing an EMR or image through ClinicStation, and I wish to document a finding, may I do that through ClinicStation or will I need to access that legacy system directly? Will that finding, if entered through ClinicStation, be available in downstream systems or merely from within ClinicStation?
As for the SOA model, I would encourage you gentle readers to take a look at an article with a not so gentle title: http://it.toolbox.com/blogs/madgreek/did-soa-die-or-do-we-just-suck-at-architecture-29157
Certainly Dr Vogel and his team have accomplished something quite significant, but I am wondering whether the article, by intent or omission, oversells what was done.
Still not getting it. In my experience, when a doc orders a test, they either get a response or stop ordering from that lab. They don’t generally have a 50% loss rate. Maybe athena just doesn’t get one electronically 50% of the time and the doc deals with it on paper. Could at least understand that.
I don’t get what Jonathan Bush was talking about either. And the explanation given above sounds like what any EHR should do as standard practice.
Am I the only one that is sick of Jonathan Bush? Maybe it’s my dislike for the Bush family and Harvard MBAs.
Fred Friendly Says: “I didn’t note anywhere in the article that ClinicStation is in fact a “clinical information retrieval system.” Unless I am mistaken, ClinicStation is a SOAP portal that allows clinicians to view data from existing systems…”
You are mistaken FF.
One fact to check on the UCI Anesthesiology software story is whether or not in that system is actually up and running and actually used…
I think people are rediscovering that yes, you can build a CDR. A pretty cool one too. Right after that is when they’ll discover that clinical transactional systems are what’s needed to drive data for a CDR and then they hit the brakes, throw away millions in sunk cost and start to look beyond their walls at commercial systems.
As most can appreciate, a paragraph or two really is really insufficient to describe an architecture and its implementation. Even case studies tend to freeze ongoing development activities at a single point time. Thanks for the reference to Mike Kavis’ blog—many excellent points are made not only about SOA, but about architecture in general. The observation that we began with SOAP-enabled viewing of data in legacy systems is correct—but that was several years ago. While providing services wrappers for legacy systems still needs to be done in some cases, we have found that our efforts have turned much more toward the actual development of services that can be re-used as building blocks for business-driven applications. Part of this involves the development of more specialized applications (ClinicStation was actually preceded by RadStation, and we now have PathStation and ResearchStation, and because of the importance of tissue management in cancer research and care, TissueStation as well). And we have become much more focused on building clinical transaction capabilities, which are certainly a challenge but don’t have us “hitting the brakes” at all. Our commitment to a services architecture allows us to view clinical transactions as sets of reusable components, with data “exposed” and “consumed” as needed. So SOA architecture benefits us from the 50,000 level all the way down to individual transactions.
But in the long run, we consider our most important objective to be the building of the linkage between the data our patients generate from their participation in research and the clinical outcomes they produce from clinical care. This is where “personalized medicine” promises to change (and in some cases has already changed) medical research and practice in fundamental ways, and we do feel that SOA offers a better architectural model for actually “getting there”.
Lynn Vogel writes: “But the cost of a major architecture change is simply prohibitive for commercial healthcare IT vendors.”
They have their feet in cement and will simply not be able to react to a new, disruptive technology. The opportunity is tremendous to the newcomer with a new approach.
Some elements of a new approach should be: 1) Very low-cost compared to current competitors, 2) non-disruptive, 3) Customizable by user, 3) easy to change and adapt to an ever-changing healthcare landscape……. I believe the technology that will support this approach is SOA as used by MD Anderson.
No doubt, MDA is pushing the envelope for an SOA based clinical system.
With 100 internal and external folks working on this, it is still a long way from meeting the 100 plus CPR criteria first laid out by IOM. Achieving effective CPOE with decision support between a foreign Rx system and yet to be developed true OE system (not just e-routing of on-line order sheets) is almost impossible given the lack of semantic underpinning (ie where is the controlled medical vocabulary?). Ask Classen.
No doubt that eventually all will be overcome. But at what cost and why. Your burn rate is $10M/yr. You’re 4 yrs in and perhaps halfway done. And if key developers/architects leave??
Good time for us all to re-visit that comprehensive IOM document and understand what truly constitutes an CPR/EMR
Lindy – how much would they be spending on a vendor system and potentially not getting the functions they want, paying for those they don’t need and wasting all their past investment in their best of breed systems?
With a team of 100, I am sure there are more than a few system architects.
CPOE with a foreign Rx system can be accomplished if you base the design on the foreign systems’ data structures, definitions and rules. Not ideal, but a solution.
Art – Let’s cut to the chase. Very beneficial that they built an SOA web viewer into all the clinical results from all their 30 odd best of breed systems. That’s a given.
However, I would think their number one patient safety/quality challenge is the thousands of high risk chemo doses they administer daily (still a paper based manual process). Granted, CPOE with decision support for complex regimen based chemo is not easy and a challenge for even the big boys (Eclipsys, Cerner, Epic). Nonetheless, they have yet to tackle this. It took Partners’ Dana Farber nearly 10 yrs to refine the famous home grown already there BICS to do Chemo OE. Memorial Sloan Kettering went commercial and has Chemo OE under Sunrise DM
Lindy and Art,
MD Anderson’s architecture is the best of bothl worlds. It does not require them to build everything. The virtual database, SOA framework they are using (developed, by the way, by Chuck Suitor) gives them great flexibility to integrate commercial systems or build it themselves. Functionality unique to MD Anderson can be built in-house and functionality common in the industry can be purchased and easily integrated.
Is that such a bad thing?