Lyman Garniss is project director for Partners Enterprise Pathology at Partners HealthCare of Boston, MA. He was interviewed by Lorre Wisham of HIStalk
Tell me about yourself and the organization.
I’ve been here at MGH and Partners for 25 years this month, primarily working with the clinical laboratories and anatomic pathology folks at the MGH. However, in the past year, I’ve been working with Partners HealthCare, the larger system, to roll out Sunquest lab and some of the blood bank systems and other products across all of Partners and also integrating that with Epic.
Epic is going in across all of Partners, and everyone’s getting a standardized HIS. We’re also at the same time moving everyone to a standardized LIS.
What are your thoughts about the future of genetics and genomics in terms of data and whether it comes from the HIS or the LIS?
The LIS really needs to own that space. It’s where the data is generated today. At least 70 percent of the data in a patient’s medical record in the HIS is coming from pathology, from the labs, from anatomic pathology, and blood bank. There’s some 10 billion lab tests that are performed in the US each year. We produce a ton of that data.
It’s going to get much larger with the genetic information, proteomics, and all the variant information that goes along with those data points. There’s no easy way to manage that today. The lab owns the specimens and some of the reporting of that, but they really need to own the data itself, the mining of that data, and the curation of the information related to the genetic variance and the proteomic data.
You’re saying lab needs to own it, which suggests they don’t now. What has to happen?
It’s a collaboration of the LIS vendors, companies like InterSystems or IBM that manage large data sets well, as well folks from the academic and medical area. A three-way collaboration that has to happen.
The lab information vendor knows the lab space really well and the lab processing and the specimen processing, but their niche really isn’t large data sets and mining large data sets. That’s where they’ll need some help from some of the large database companies, as well as some input and advice and boots on the ground from some of the folks that are actually performing these tests.
You mentioned Epic. What do commercial electronic medical record systems have to do long term to keep up with medicine on the lab side?
The breadth and scope of HIS vendors are so large that it’s very difficult, if not impossible, for them to be experts or to be best of-breed in every single area or every single domain that they touch. They may have some expertise in specific areas, but they would have to invest a lot of money in the lab space and really be focused on that. It’s difficult for large vendors like Cerner and Siemens and Epic to be experts at the lab and also manage that large breadth and scope. Their domain is huge.
Why are you not choosing to implement Epic’s Beaker LIS instead of Sunquest?
There’s a number of reasons for that. The first one is the Beaker product just doesn’t seem mature enough for our needs. That’s one data point. The second data point is we have a rich history of collaboration with Sunquest. We’re working on integration between the laboratory system and the anatomic pathology system.
The walls between AP and CP are breaking down. The type of work that is happening in the lab information system is starting to look, in some instances, more like anatomic pathology — large, rich textual reports that the CP system doesn’t do well but AP does well. The reverse is true. We’re doing more instrumentation and instrument testing on the anatomic pathology side. The AP systems really don’t manage instrumentation and stuff well, where the lab system does.
Sunquest is moving forward with us on breaking those walls down so that the integration between AP and LIS, the APLS, and the LIS is much more robust and streamlined, much more integrated.
What do you think a hospital CIO should know about lab or pathology informatics?
I was invited to speak to several different groups of my peers here at Partners.The message that I was trying to get across to folks in IS and to other areas is that, again, we’re providing 70 to 75 percent of the data that’s in that HIS. The lab information system is large, it’s complex, and we’re producing a lot of data, a lot of rich data. We’re going to continue to produce even more of it in the future.
How are you using your EMR for clinical decision support to guide physicians in ordering labs?
Poorly. [laughs] We have a current provider order entry system. It’s a homegrown system. We are able, with that homegrown system, to customize it to some degree to help steer clinician ordering away from expensive send-out tests.
I’ll give you an example. We show dollar signs next to the test. We don’t actually say how much it costs, but we provide alternative tests that would be cheaper to keep the costs down, especially as we move towards that ACO market. We’re starting to look on the lab side how to steer clinicians towards tests that may be just as effective, but may be less expensive. Epic doesn’t really do that today.
Clinical analytics for population health management is a hot topic. What is Partners doing to standardized data and enable better reporting?
The first step is moving to Epic. I believe that the folks at Partners are working with Epic to work more on some of those outcome models and patient care models moving forward.
How do you see personalized medicine based on patient genomics moving into everyday practice?
It has to start with the discrete data that’s in the laboratory. We can’t possibly send all the genetic information to the HIS. It would be overwhelming for the clinicians. It would be overwhelming for the HIS system itself.
The role of the LIS vendor is going to be looking at all these rich data sets and mining, looking for patterns for outcomes to figure out both on the CP side for predicting potential disease states based on genetic variance, but also on the anatomic pathology side, looking at outcomes and survivability for specific types of cancer to figure out.
It’s probably best if I use an example. Something like pancreatic cancer, where the survival rates of that are horrific. Once you’re diagnosed with pancreatic cancer, usually you have weeks or months to live. But there’s probably people that survive longer than others, and we have their specimens. We have a rich biorepository of anatomic pathology specimens, where we can look at what’s the genetic variance in a particular type of that outcome versus somebody that didn’t survive as long. Then we can start targeting therapies based on that.
The example that everyone’s using today is the HER2. There were one or two treatments for breast cancer in women. Depending upon the treatment, sometimes it killed the patient quicker, but sometimes it cured them. But there was no rhyme or reason until folks figured out that we could base the therapy on specific genetic markers in the tumors.
HER2 is a perfect example and more and more of those examples are going to come forward. But the only way to find those models or find those differences is to be able to mine the data. The laboratory owns that data. It doesn’t reside in the HIS. Mining of that data has to happen in the LIS space.
How can the average hospital involve the lab people in their EMR decisions and setup?
I have attended the Sunquest User Group for many years. There’s now Epic roundtables at SUG, the Sunquest User Group. Those have been very enlightening.
What we’ve seen at those meetings are basically two camps of folks. There’s folks that Epic was installed at their site and the laboratory was not involved. In speaking with those lab folks now, Epic has become a nightmare for them because they were not involved with a lot of the decision-making process. They don’t have access to a lot of the way that the system was designed or built. Supporting the lab information system and integration with Epic has been extremely problematic for those sites where lab was not involved up front and early with the HIS vendor with that integration and testing process.
Then there’s the other camp, where the LIS folks — both IT and the actual people that are doing the science in the laboratories — were involved early on with the HIS implementation. Those folks are happy with Epic. They’re happy with their HIS and things are going extremely well. We learned from that at Partners early on. My team and others on the lab side are working really closely with the Epic folks in installing the systems here and doing the integration work. It’s been absolutely fabulous. A lot of decisions are being made that make sense for both sides of the equation. Again, we’re providing 70 to 75 percent of the data that’s in the HIS. The lab has to be involved early on with the design, setup, and decision-making process that goes on with the HIS build.
How is Partners exchanging information with other facilities outside of your group?
We’re part of the Mass HIway. That interchange is still being rolled out. I wouldn’t say it’s in its infancy, but it’s in its adolescence.
We send quite a bit of information to the Mass HIway. We do all of our state and city reporting to the Mass HIway. Data that has to go to the state for state-required reporting, like some of the blood management things and the microbe reporting.
Partners HealthCare provided the seed money for an exchange years ago called NEHI. It was for medical centers in the New England area to exchange information with the insurance companies. We’ve been doing that for years. Instead of just having the Partners facilities manage that, we invited other institutions as well because we thought it would be in everyone’s best interest to share the cost of that insurance exchange with them.
What are your biggest challenges and opportunities at Partners over the next year?
There’s a few things that we’re working on. Expanding the LIS into areas that they typically haven’t worked in. One is research and managing the research specimen flow; the second is biorepositories. There’s a lot of rich specimens flowing through our systems, but we’re not able to track cohorts or manage consented patients in the LIS, so we don’t know when their specimens are flowing through the system to be able to move those specimens elsewhere or inform a researcher that specific, unique specimens are available. We’re throwing tons of these specimens away and they’re actually quite valuable in some instances. The biorepository area is something that we’re working with the LIS vendors on and we’ll expand in the future.
And then of course the genetics, genomics, and proteomics, and the rich data set … curating the data that goes along with those. The variant information. It’s a huge challenge for all of the institutions that are doing genetic testing to be able to track and update the most recent information about specific genes or gene variants. It’s a huge challenge for folks. There’s no straightforward tool sets that manage that rich data set today. It’s one of those things that the lab, the LIS needs to own and need to expand their role in.