Memorial Hospital at Gulfport, MS, went live with a new Cerner EHR in June 2014. Before the “bigbang” implementation, the hospital laid the groundwork for using the EHR for clinical and management improvement by building an extensive data warehouse and analytics framework with technology from Health Catalyst, a Salt Lake City, UT-based company that spun out from Intermountain Healthcare.
Here, Gene Thomas, VP of information systems and CIO at Memorial, discusses how the enterprise data warehouse works in tandem with the EHR and offers advice on how to build and implement such a warehouse.
Compiled by Neil Versel, Contributing Writer
Memorial Hospital’s CIO shares his strategy and best practices for establishing an enterprise data warehouse that is a “single source of truth.”
Memorial Hospital at Gulfport, MS, went live with a new Cerner EHR in June 2014. Before the “bigbang” implementation, the hospital laid the groundwork for using the EHR for clinical and management improvement by building an extensive data warehouse and analytics framework with technology from Health Catalyst, a Salt Lake City, UT-based company that spun out from Intermountain Healthcare.
Here, Gene Thomas, VP of information systems and CIO at Memorial, discusses how the enterprise data warehouse works in tandem with the EHR and offers advice on how to build and implement such a warehouse.
Q: What are you doing in terms of enterprise data warehousing at Memorial?
A: We had been building best-of-breed systems for many years, as were many healthcare organizations. A couple of years ago, we strategically got centered around the idea that we wanted to have as much of an integrated system as we could. If we were going to spend the time, money, energy, and discipline for an integrated system for data collection and providing care, we also wanted to have robust enterprise data warehouse capabilities, as well as the analytics you could lay on top of that data.
Q: How did the EHR rollout go and how did you tie it to analytics?
A: About 2½ years ago we did a vendor analysis, brought in four finalists, narrowed the candidates down to Epic and Cerner, and ultimately chose Cerner for our very large, bigbang, housewide EHR replacement. We also picked PeopleSoft for our financials at the same time. In that same project, we put in, budgeted for, and got approval for a robust enterprise data warehouse and analytic capabilities. We ultimately chose Health Catalyst.
We went live with Cerner on June 14, 2014. We actually built an enterprise data warehouse in advance of going live with Cerner, and we brought all the relevant historical clinical information from our legacy EHR systems. They were disparate — one for inpatient, one for ambulatory, one for the emergency department (ED), and one for NICU (neonatal intensive care unit). The day we went live with Cerner, we had a daily feed that populates our enterprise data warehouse.
Q: As for the analytics, what sorts of metrics did you decide to dive deeper into?
A: It’s a long laundry list of metrics. We want to look at cost per case, length of stay, variability, utilization, associate quality, and outcome components. Additionally, we look at patient severity indexes and try to map those to everything from what treatments are used to what antibiotics are prescribed, and what staffing skill level is required. These are all the things people want to look at to help drive — not make, but help drive — better decision-making.
Q: You’re talking about both managerial and clinical decisions?
A: Yes. We believe in a healthcare delivery system that is integrated, and all of those things need to be analyzed so you can begin to discern what decisions make sense for the type of patients you have and your procedural areas, whether it’s medsurg, radiology, pediatrics, adult, or cardiovascular.
Q: Was there anything you found surprising with regard to your areas of strength or weakness?
A: We previously had analytics in spots, but we didn’t have the ambulatory component married to our inpatient system. They were two disparate systems from two different vendors. We could do a little bit of analytics on our inpatient population and a little bit of analytics on our ambulatory footprint but, of course, some of those patients are common to both settings.
We’re now able to tie those together. We’re able to find, in some cases, early on, potential cause-and-effect relationships. We can see where patients have been compliant just because they’re showing up for appointments or not. We found areas where there are potentially some care processes where we could tighten up variability.
We want to get the data, get the analytics, and make sure that the data is accurate and can be trusted. We’re beginning the process of being able for the first time to give our physicians, our clinical staff, and our quality and safety people holistic data enterprisewide on a patient or a population of patients. Once you’ve got that view, you can decide what care plans you want to modify. Without that cross-enterprise view, it’s difficult to do.
Q: Memorial was in the path of Hurricane Katrina 10 years ago. How did that disaster impact your approach to data management?
A: New Orleans got the press, but the direct hit was here on the Gulf Coast. A lot has changed since then, and not only with our system. After Hurricane Katrina, many patients went away and a lot of practices went away with them. When I say the practices went away, either the patients went away or the providers’ buildings went away, or both.
Our mission as a community-based hospital is to be there for the community for care. Three or four weeks after Hurricane Katrina, the administration (in conjunction with the board) offered every one of our affiliated community physicians employment with the hospital. Very few physicians took us up on that offer initially, but over time, more and more did. We now have a footprint of somewhere north of 83 clinics that are owned by the institution. We’re more like a health system now, not just a hospital, and that led to a change in our data strategy.
After Katrina, we started to grow that ambulatory footprint, and we started to collect data in the ambulatory space as well as in the inpatient space. We automated the ED, so we’re collecting data there. As I looked at Meaningful Use and tried to figure out what we’re going to do with population health and care coordination, we decided to stop investing money in our disparate legacy systems. We wanted to look at a holistic, integrated system.
We have all this robust data now, so we can actually do something meaningful with it from an analytics standpoint. We just didn’t have a way to tie it together with those disparate systems.
There has got to be a way — and I think we can be one of the proving grounds — to put in the right systems and get the right data. You still have to rely on the physicians and the clinicians and the nursing staff to provide the care. But give them the data so we can figure out how to lower costs and manage a population.
Q: What is your approach to data governance at Memorial?
A: When you build an enterprise data warehouse and hospital leadership endorses it, all of a sudden the requests come out of the woodwork. There has got to be discipline around what you do with data warehousing and analytics. You have to start by establishing some foundational elements such as data governance, common definitions, and policies and procedures. Staff education is also key.
If you don’t do those things right, you could potentially wind up with some erroneous data and some data variability, or the way people define the data or view the data could be less than ideal. The sooner we get over that hump, the better off we’re going to be. The sooner you get there with credible data — and underscore “credible” — the better off you’re going to be.
My job is always to make sure the data has integrity — that it’s a single source for truth, so our clinicians can make better decisions. I consider the IT department to be a service organization to the medical staff, the quality department, and the safety department.