Reproduction permitted for personal use only. For reprints and reprint permission, contact email@example.com.
- As the director of applied sciences and CIO of Marshfield Clinic, Dr. Robert A. Carlson is in an enviable position. Not only does he serve an organization with unquestioned leadership in healthcare information technology, he has the luxury of focusing on strategy (thanks to the presence of chief technical officer Carl Christensen and chief medical informatics officer John Melski, MD.) and the expertise to bring a physician's perspective to the development of healthcare IT. All of which will come in handy as Marshfield Clinic continues to drive HIT advances through programs like the Wisconsin Gemomics Initiative.In this first part of a two-part CIO Leadership interview, Carlson (a University of Wisconsin-Medical School graduate) talks about the next evolution of HIT at an organization that was among the first to adopt electronic medical records and has commercialized them in a product called CattailsMD.WTN
: Marshfield Clinic has its technology management positions divided three ways: Carl Christensen is the chief technology officer; John Melski is the chief medical information officer; you're now the chief information officer. Explain to me the division of duties and whether this arrangement frees you to be strictly strategic on matters like the genomics project with UW-Madison, UW-Milwaukee
, and the Medical College of Wisconsin
: Carl Christensen fulfills the chief technical officer position. That's the love of his career life and really plays to his strength. The reason that we split that off was really at Carl's request. He saw the changes coming down and the need to start really focusing on the technology.
The other part was that the clinic really wanted a physician point on their IT infrastructure. The obvious choice would have been John Melski, who is currently the chief medical informatics officer. I think [he passed] partly because the other part of his life is being the dermatology chair, and partly the issue that was really defined is how do you start facilitating and plugging the practice of medicine, and more importantly where medicine needs to go, into the IT structure?
So in my role as CIO, I have the luxury of having both Carl as the chief technical officer and John Melski as the chief medical informatics officer. I can really focus on the utilization of our informatics resource and how it needs to evolve and facilitate the effective and efficient practice of medicine. A lot of my time is plugged in with clinic leadership, with physicians around this system and systems that use our software, to really help understand where medicine is going and bring to bear the tools that IT has to offer. WTN
: Does this allow you to be more strategic on matters like the Wisconsin Genomics Initiative?Carlson
: Yes, I am way off the scale on the strategic end on this one. WTN
: Then your primary role is to align healthcare IT with the practice of medicine?Carson
: If you look at what we have here at Marshfield, we have a very large group practice of medicine with a long history of using computers and information. We have the personalized medicine [genomics] project and the largest bio bank in the country. We have one of two or three bioinformatics research centers. We have a practice of medicine-computer relationship that's to die for. I feel like I've got a Ferrari
. My problem is that it's in one thousand pieces on the garage floor. So how do you start to align these pieces and how do you get them to a point not to where medicine is today, but where medicine is going in the very near future.WTN
: Is Carl pretty much the point man on IT implementations, or do you still play a lead role in that?Carlson
: It's the strategic part that I really play a heavy role in. When we start to get into the implementation piece, you begin to hand off more of the technical aspects of that. The bottom line is Carl's office is right across the hall from mine. We need to talk frequently. If you look at what I do and then what Carl does, there is not only an overlap but also a lot of iterative communication that goes back and forth.
What I'm doing now is rethinking our work group structure. How do I integrate with the physicians? One of the things we're dealing with right now is with Ministry
[Health Care] as a client of Cattails, we're really sharing a common medical record. There are a lot of issues that you don't think about on the front end that start falling out when you start to implement things like standardization of information.
I'm meeting with physicians and I'm fleshing out these unanticipated problems that come up, looking for stakeholders to take ownership of these problems, and building committee structures to get those problems adjudicated. Then quickly I'll be getting into where we need to be going and helping align vision, and helping to make sure the strategy of the Marshfield Clinic is articulated in a way that we then can provide direction of resources within IT.
As that discussion starts to become more technical, when we start to talk about capacity sizing, the number of cycles that you've got to go through to get something done, or long-term development, now you're starting to drift into things that Carl would engage on. WTN
: We've all heard the stories, fictional or otherwise, about how doctors haven't exactly taken to electronic medical records and other healthcare IT like a duck takes to water. Obviously, that is not an option at Marshfield Clinic and you've tied the record to the practice of medicine, so how will your background as a physician help you as a thought leader in the technology realm? Carlson
: When you look at the [clinic] appointment system, it's based on the fact that when you get sick, it's geared around patients pulling the cord and engaging in the healthcare system. As we start thinking about how you effectively manage patients like diabetics, waiting for someone to end up in the ER and having a three-day hospital stay is not how you want this to go. So how do you identify at-risk patients and go out and interact with them to prevent
them from using the ER? Where I'm engaging in is trying to say the appointment system is reactive. What we need is a proactive way that maybe ties into the appointment system. That's one piece.
The other thing I do is go out and meet with physicians. I can tell you right now that one of the struggles that doctors are having is that more and more of their practices are now being required just to get the billing coding right. A lot of their time, whether it's the reimbursement system of clinics as a whole or the salary structures for physicians, is being spent trying to get the coding right, the billing right, and making sure you've documented the necessary levels of care. Less and less time is available not only to spend with patients, but to really get at the heart and soul of why you're there, and that's what do you think is wrong and what do I need to do to get it fixed?
So are there ways that we can embed into the electronic record these intelligent assistants that, as they watch the process unfold electronically, bring the right information and the right decisions and the right nodes and the right helpful suggestions to make them not only more efficient, but allow them to spend their time and energy on where they need to spend it - on what is wrong with this patients and what do we need to do with this patient? WTN
: Is that the next evolution of the record?Carlson
: I think it starts out as being kind of a virtualization of the paper world, but I think Carl Christiansen said it right - an EMR is just a race to the starting line. All of the sudden you have the ability to bring to bear what other industries have been doing for quite some time, and that is this intelligent assistant kind of approach. This includes things like helping sort through a differential diagnosis. One of the things I think we've just begun to tap is the whole field of healthcare intelligence, where you take the analytics and the business intelligence tools and start applying them to the field of medicine.
My daughter's boyfriend has an iPhone
. There was a song playing in the background, and he called up an application on his iPhone and it listened to this song, recognized what song it was, and put it out to iTunes. And it was there it was for him to purchase. If we can do that with music, why can't we do that with EKGs? How do you take that kind of intelligence and now apply it to medicine?WTN
: A while back you mentioned that Cattails would need to be modified to accommodate Marshfield's role in the Wisconsin Gemomics Initiative. How so?Carlson
: I can't get into all the specifics because that isn't necessarily my expertise and it's an evolving field, but the ability and the privacy and the implications of genetic information are much broader than a simple blood test on a chemistry value. There is so much more information in a genetic test, or a genetic sequence in terms of unintended questions or unintended answers to questions that you didn't ask. If you have a risk for a certain disease, what do you do with that?
So I think the rules are still being defined on how genetic information needs to be protected, how it needs to be stored, and its association with that [record]. That being said, if you can broaden the question out into more of a data mining or an analytics approach, where you think of the question, Why is one patient's outcome better than another's?" Certainly genetics plays into that, but there may be other things like environment and those kinds of things.
The key to that is going to be the accuracy of your data points - for example, the diagnosis of diabetes. Diabetes is a laboratory-driven diagnosis and yet if you're searching on the diagnostic term diabetes mellitus
, what we've found is that there are references in the chart that don't meet the criteria for being diagnosed per se as diabetes. Yet if you're trying to sort through and mine outcomes, and you're measuring and you're counting as your anchor point a diagnostic term that's not accurate, you quickly lose your ability to find those relationships.
So one of our efforts is to really go back and make sure those data elements that we're going to be mining on - or trying to derive relationships on in terms of outcomes - that we really have to spend a lot of time understanding what the criteria are, and when we put those data points into a data mining field that you're going to have sufficient accuracy to make it worth your while.Related story
Recent CIO Leadership Profiles
In Part II, Carlson further addresses the evolution of Marshfield Clinic's electronic medical record, the accuracy of EMRs as a potential pitfall in applying business intelligence in the healthcare realm, and the impact of economic conditions on HIT deployment.The Robert Carlson bio:
- Director of applied sciences and chief information officer, Marshfield Clinic; committee chair of Marshfield Clinic's Technology Transfer Advisory Committee.
- Graduate, University of Wisconsin Medical School, where he also completed a pathology residency.
- Diplomate, The American Board of Pathology in anatomic and clinical pathology.
- Marshfield Clinic, 1987 to present; has served as chairman of the department of pathology, medical director of corporate communications, section head of clinical chemistry, section head of the reference laboratory, and director of the Division of Laboratory Medicine. (Responsibilities have included oversight of clinical laboratory services, coordination of molecular diagnostic services and DNA storage in the Personalized Medicine Research Project).
- President of Marshfield Food Safety, LLC.
- President of the Wisconsin Security Research Consortium.
- Executive board member of Wisconsin Technology Council.