I read several articles about Berkshire Health Systems, Pittsfield, MA and their work with CPOE and other clinical systems. Michael Blackman is the CMIO there and is frequently quoted about clinical systems implementation and the value of CPOE. He agreed to be interviewed for HIStalk, which I appreciate.
Tell me about yourself and your job.
I’m an internist and med-peds trained. I’ve been at Berkshire pushing seven years at this point and have been CMIO for one. It was a new job for this organization.Berkshire Health Systems is two hospitals. The larger one is just over 300 beds and a teaching hospital. We have a few small residencies in medicine, surgery, pathology, and soon to be psychiatry, starting in the summer. Those are all UMass-affiliated programs. The smaller hospital is a critical care access hospital with 27 beds, 25 or 30 miles south of us. We also have 15 owned physician practices and 2,000 long term care beds in Massachusetts and some in Pennsylvania and Ohio.
We’re putting in a completely integrated record across that entire enterprise.
Did that project lead Berkshire to create the CMIO position?
We were doing so much in the space is that we needed more physician time than we had before. We needed a physician champion with a focus on getting the last pieces done.
What systems are you using and what projects are underway?
For projects, we have CPOE up and about 75% of our orders go through it. The goal is to push that to 100%. What we don’t have is eMAR, the electronic meds Kardex. It’s been a holdup for us getting CPOE pushed the rest of the way. We bought the systems up in a strange order, frankly, in part because of what was available at the time. We’ve had CPOE up in one form or another for almost four years.
Other big projects are getting the ED on CPOE and getting all of our outpatient practices up on the electronic record so they fully integrate with the hospital. Those are the biggest things at the moment. Also BMV – closed loop verification. How did I forget that one? It goes with the eMAR.
You’re all MEDITECH, right?
We are all MEDITECH.
How would you evaluate your success with it?
It works. I happen to like MEDITECH as a system. The pieces actually all really do talk to each other. The integration, certainly for an organization this size, is their problem, not ours. They make it work. We don’t worry about an upgrade and making sure that the orders from CPOE are still going to pharmacy the right way. We know that’s going to work and that makes a very big difference.
Most of your doctors are community-based, I assume?
Most are, but the hospital employs probably 80 to 90 physicians. That includes ED, outpatient, and hospitalists.
Is the majority of the 75% CPOE ordering coming from those employed physicians?
At the moment, it’s really the house staff and employees of the hospital. The reason is that we haven’t pushed it to anyone else. We’ve being waiting until we do eMAR. We definitely have some private physicians who use CPOE, but it’s them coming to us and saying, “We seen this and we really want to do it.” It hasn’t been me going out and saying, “OK, now you’ve got to get on.” We want to be sure they can go to one place and get everything they need before we do that. We want to get eMAR up and then we’ll take that step.
What is it they like well enough about CPOE to want to volunteer?
Some of them look at the hospitalists and say, “Wait a minute. They have these nice order sets they can just click through them and get their stuff in,” especially for specialties where the orders are fairly routine and standard. Orthopedics, for example, or OB. Admitting a pregnant woman for routine delivery is essentially the same every time you do it, so people that were really using order sets or felt they could … those are the groups that have stepped forward to say they want to do it. It will get trickier at the margins.
The biggest challenge is always ED, ICU, and surgery. Will they like MEDITECH?
I think so. As is true with any system, none of them do everything you want them to do. The upgrade is always hopefully better than what you had. Even then, you find features, so, “Now that it can this, I want this as well.” While we’ve done rather well with the CPOE we have, there have been moments when we said, “We really need what’s in the next version.”
Our next upgrade will help us a lot in that regard, most notably in our ability to convey more context to people as they go through order sets. You think about order sets on paper – we use them a lot for teaching … “Here are the criteria for this.” It’s hard in our current system to present that information in conjunction with the order sets. The next version makes it quite possible to do that.
Is there enough CPOE value other than efficiency with order sets?
Order sets are faster than paper, but they’re still attractive for decision support, even if just allergy and drug-drug interaction and formulary compliance issues and cutting down on the calls from nursing and pharmacy to clarify orders. We don’t try to sell it because it to people as, “You should use this because it will make you more efficient” because we’re going to be wrong. It doesn’t always make you faster. It may make it safer or better, but not always faster. I think a large number of the vendors have done all of us a disservice by touting the fact that they think this thing is faster and it isn’t, always.What clinical decision support capabilities do you have turned on compared to when you first brought up CPOE? Most hospitals dial it back.
If you look at what we did originally, we said, “We have CPOE now and we should turn on all the decision support we have available.” Boy, was that a mistake. [laughs] We turned most of it off rather quickly. At this point in time, we have drug interaction checking, allergy checking at both the generic and ingredient level, and adverse drug reactions by class. We have a variety of rules around lab utilization, formulary compliance issues, and some financial rules.
We found that we were losing revenue from visits we could not bill for because there was no admit order on the chart. Somehow it got overlooked. They came up from the ER with holding orders and a continuation of those orders, but there was actually no order on the chart that said, “Admit to blank.” Without that, the payors were denying the bills. One of our rules for an inpatient is that you have to have an admit order before you can order medications or labs.
What devices are the physicians using?
People either use tablets or desktops. One of the issues we ran into as we rolled out tablets to the employed physicians is the lack of flat space to put them down on the units. People had a tendency to put them in front of their existing computer, which wasn’t the point. We were trying to free up computers for the nursing staff.
Did you encourage broad order sets but not personal order sets to avoid future maintenance and to reduce variation?
The process was really trying to keep them as evidence-based as possible using our existing structure of department-level order sets. We do currently allow people to save favorite sets. We don’t teach them how to do that, but a few have figured it out on their own. We’re looking at changing that as we move forward because the maintenance around personal order sets is an absolute nightmare.
On top of that, if you have evidence-based sets and you need to make a change in the drugs for some reason because of a change in the evidence or change in availability, we want to be able to push that out without worrying that some sets missed it. Depending on how people save their favorites, whether they built them on existing ones, that might or might not happen.
Have you done any metrics to determine what has changed since CPOE?
The good news is that our medication error rate has decreased. We’ve seen a definite drop in the med error rate every time we bring another group of users on CPOE. We’re looking at those errors and anxiously looking forward to how that changes when we get to eMAR and BMV. We think that will eliminate a good chunk of the remaining ones, with the goal of zero.
The incidental thing is that we saw an increase in the number of duplicate orders that had to be canceled by the lab. One of the things we didn’t pay enough attention to was, “What were we taking out of the system?” We took out the unit secretary from entering the lab orders and that human intervention of saying, “There’s already an order out for this” and how the lab functions. I’m not sure we weren’t writing as many duplicate orders, but they were just getting through to the lab. We put some rules in place to try to cut back on that and we’re about to re-study that.
I assume your biggest impact will be when you get BMV online. What challenges do you see with that?
We’re doing the eMAR and BMV together. We felt the workflow for nurses wasn’t appreciably different bringing them up together since we were training all of them at once and some of the efficiency came from BMV.We have a barcode packager, so everything has to go up to the floors in unit dose. We’ll have to repackage anything that doesn’t come in unit dose. We’re doing our best to keep the formulary as standard as possible. The other piece is the workflow one – making sure we check once a day or once a shift to make sure people are actually using the scanners and that they’re scanning the patient’s wristband and not some list on the wall.
Are checkguards against that built in?
We’re probably going to put a check digit on the wristbands or some of the labels so you won’t be able to interchange them.
Do you have informatics folks working with nurses?
We have a variety of people working with nursing, but it’s still me, going back and talking about some of these things. On our IT staff, we have several nurses who function as both analysts and liaisons to nursing. Plus, we tend to pull people in for these various projects from the respective departments and then backfill on the units.
What’s the plan for EMRs for the practices?
We’re going to use LSS, which is the MEDITECH-affiliated outpatient product, so we can get the real value of the seamless integration between the offices and the hospital. We have all of them up on billing and scheduling and it’s been that way for a couple of years. Now we’re ready to move forward in pursuing the clinical pieces so that if we make a change to the meds list in the office, we see it in the hospital and vice versa.
What’s the level of integration for allergies and updated information?
The answer is that it’s 100%. It’s the same allergy list, the same outpatient med list, the same problem list.
Does it handle medication reconciliation?
That’s something they’re still working on. We’ve done some things for med rec on the hospital side, mostly building reports that compare an electronically entered home medication list against the admission orders. We still drop the paper and the computers are lined up what’s a match and what’s not and someone reviews that to see what they want to continue and that the changes are correct. We have a physician and nurse review and sign it.But the biggest piece is that they look at that piece of paper and say, “This isn’t what I meant to do.” They don’t make the changes on paper, they go back and make the changes in the system and then reprint the report. We hope to eventually do all that electronically, but it’s not quite there yet.
Are you doing any interoperability projects?
I don’t know if you’re familiar with the eHealth Collaborative in Massachusetts, which was three pilot sites to do interoperability in the community. North Adams is 20 minutes north of us and they’ve done some things there and we’ve had discussion, but it’s not there yet what we can do to integrate the county. We haven’t done it, but we’ve definitely talked about it.
There’s a lot of MEDITECH in your part of the world, which should make it easier to share data.
Massachusetts is MEDITECH central. I think they have 70% of the market in Massachusetts or something like that.How well would you say that today’s systems align with contemporary medical practice?
One of the biggest struggles and the last piece for most people is physician documentation online. We haven’t done that yet, either.If you listen to the vendors and listen to the industry, there’s a big push to get as much stuff in structured data fields as possible. That’s great for reporting information, but it’s not how physicians tend to think. It also doesn’t lead to notes that really tell you what you want them to tell you.
If you take the structured notes at one end of the spectrum and the complete, dictated, full-text note at the other end of the spectrum … the other good piece about the structured ones is that they’re better for billing, but the free-text ones really tell you more about what’s going on the with the patient. I think the right answer is probably somewhere in between in making a note that gives you a good combination of structure where it’s appropriate and where you really need it for reporting, and allowing the free text to give you the full flavor of what the patient looked like so someone subsequently reading it can tell what happened. Often it’s fully templated and you can’t tell that.
The article that just came out in NEJM said that the urge to get information into a template hasn’t done the patient any favors because there’s no context.
The other piece about the NEJM article is that they said that people were committing clinical plagiarism, cutting and pasting large sections of notes that aren’t theirs without really reading them. This is a tool like anything else. You want to be able to pull certain things forward, but equally you want people to use the information effectively.
The best example I can give isn’t mine originally. You can pull enough information from the system that can create a template that starts with, “A 23-year-old female presents with left ankle pain.” You’ve got the age, the sex, the chief complaint, but there’s not a template in the world that changes that to, “A 23-year-old female Olympic figure skater presents with left ankle pain.” Those three words have dramatically changed the flavor of the patient and what you’re going to need to do and what the patient’s rehab course will look like. That little bit will make a tremendous difference to how people get their work done.
The other piece in this part of the world and the West Coast as well is pay-for-performance, PQRI, and a variety of other things that require capturing information in a way that’s reportable unless you have an army of people doing nothing but chart extraction.
Do you see any technologies coming up that could be useful?
It’s nothing in particular, but it’s mostly the things that focus on workflow. One of the things that a good chunk of systems today is that the workflow of how people get things done and think about things is sometimes missed. I’m not trying to suggest that you should simply take the current paper process and make it electronic, but there’s a thought process that goes about how you approach a patient and gather information together.
Perhaps looking at that better at work flows through screens … I actually had an IT physician we were talking about trying to reproduce a three-page, tri-fold flow sheet in the ICU. After we had a long discussion, he looked at me and said, “You know, perhaps now that we have all this information electronically, we don’t need the flow sheet in the same way we did before.”
I think that could be one of the changes – how do you look at the information differently and how do you best to reduce far more information out there than you can possible review? When everything was on paper and you had to get the five volumes of old medical records sent up, there was no expectation on anybody’s part that you actually read all five volumes. You may have peeked through to try to find what you were looking for, but nobody really expected you to look at it all.
Now that it’s all electronic and easy to get to, there is some weird belief that you’ve actually looked at it all, which you haven’t. So, how do we bring things that are important to people’s attention, the very complex decision like what do you put on the first screen to make sure people see? How do you alert them that there’s something in the system they should know about buried in that fifth volume of old charts that they never would have found before?
The other thing that will make a big difference is the electronic prescribing, especially on the outpatient basis. The lack of ability to electronic prescribe controlled substances is enormous. It has to happen. The question is what is the DEA going to require to make that happen? We’re working on a project around that now, so hopefully we’ll have some good news about that in the upcoming months.
What led you to leave medical practice and get into informatics?
I haven’t fully left the practice of medicine. I’m not doing as much clinical work as I would like, but I’m still seeing patients with the house staff and things like that. For me, it was an opportunity to make a bigger impact on how medicine is practiced as a whole and how we take care of patients as opposed to doing it one at a time. prior to medical school, I was a management consultant and a lot of that was information systems, so for me it’s turned full circle.