You just put the last screw in the enterprise EMR. And the day you're finally finished — after millions of dollars and years of 12-hour workdays — you start the process of throwing it out and getting a new one.
Though just a few years ago this scenario might have made industry headlines, today it's not that unusual, as more and more CIOs are tasked with migrating a legacy EMR to a new one. And most agree that going from EMR to EMR is more complicated than going from paper to electronic records. Looking at the cost and difficulty, why are hospitals taking on this gargantuan task? And for those that do, where do they even start?
“The question every CIO should ask is ‘what's driving the decision?’ — because that defines where you start,” says Jonathan Thompson, vice president at Minneapolis-based Healthia Consulting. “Functionality is a key driver.” Thompson says a common reason for change is keeping (and growing) market share in the community; in which case, a hospital starts with its affiliated ambulatory strategy.
For Bert Reese, CIO of Sentara Healthcare, a seven-hospital system based in Norfolk, Va., that was exactly the case. Reese had a lot at stake: six of his hospitals had been on an Atlanta-based Eclipsys EMR since 1994 — and it was working well. “I had standardization across the board,” he says. “We were a poster child for Sunrise Clinical.”
But in 2004, the president of his health system decided on CPOE as an enterprise strategy, and Reese realized that for Sentara, primary care, homecare, and the physician practices had to be part of the plan. “What we concluded was that if you have an integrated environment across all the environments of care, one plus one equals eight,” he says. So Sentara changed its strategy from doing CPOE to implementing a fully integrated environment.
Some of Sentara's physician practices were already using GE Logician as an EMR. “We could have interfaced Logician, but that's never as good as an integration,” says Reese. “I went back to my boss and said, ‘Would you be willing to go backward to go forward?’” Sentara opted to sign on for a full Epic (Verona, Wis.) implementation. “This will cost us $237 million over 10 years — with an ROI somewhere north of $35 million a year of new revenue and decreased costs as a result.”
Know thy user
For other hospitals, functionality can bring about better documentation. At CentraState Healthcare in Freehold, N.J., CIO Neal Ganguly says his clinical staff was the driver. CentraState had been a Meditech shop since 1999, and Ganguly says the pain points were with nursing documentation, the EMR itself, and physician order entry. “It depends on how demanding your users are — and our users were very demanding,” Ganguly says. Though nurse productivity got him started on the project, it really just opened the door. “When we began to look at other systems, we realized we wanted a lot deeper analytic capabilities, and a lot more ability for customization of workflow.”
By 2003, Ganguly had decided to move forward with Siemens' (Malvern, Pa.) Soarian, a system that he felt was more user friendly for nursing documentation. His decision was not all that difficult for the hospital to swallow — he had the advantage of the “new” on his side. “I was lucky in a way because Meditech had been selected by my predecessor — it had gone live two months before I came on board.” CentraState had also just gotten a new CEO and COO, “so it was the perfect storm,” says Ganguly. The new administration wanted to use technology to deliver value and, according to Ganguly, the Board heard the justification and understood.
He had another factor in his favor: cost. “Lucky for us, our Meditech system wasn't that expensive,” says Ganguly. He concedes that the $3 million dollars CentraState had already spent on Meditech was significant, “but if we had McKesson or Cerner in place instead, where we might have spent $8 million, it might have been a lot harder to justify.”
For other systems, interoperability drives the change. Sharp HealthCare, a four-hospital system in San Diego had been working on an EMR strategy for 20 years. “But in the past two years,” says CIO Bill Spooner, “we decided that the current strategy had too many systems and too many interfaces.”
Sharp was using United Kingdom-based GE for order entry and pharmacy, but had clinical documentation from San Diego-based Clinicomp that had been developed for Sharp. “Despite the fact that it was an elegant documentation system, the combination of multiple systems was not functioning well from a workflow standpoint,” says Spooner. “We decided to make this migration to fewer systems.” Two years ago, Sharp began implementing Kansas City-based Cerner's EMR.
Spooner had another very important driver, one that many CIOs are dealing with today: he was opening a new hospital, designed in 2000 to be paperless. He says it soon became clear that the EMR strategy for the new hospital had to embrace the other hospitals as well. “We realized we either had to make that leap or redesign the hospital,” says Spooner. “And there wasn't a great deal of eagerness to redesign the hospital.”
No matter what the driver for a new system, someone has to be accountable for it. As Spooner says, “Ultimately I get the credit if it doesn't work.”
Who's the boss?
So once the decision has been made to acquire a new system, and selection finalized, who's running the implementation?
According to Thompson, ownership and planning need to begin at the highest level. When it comes to the timeline and scope of the project, he says planning has to include the high-level hospital organization, the new vendor of choice, and the current vendor. “All three are stakeholders in the transition and accountable in the planning,” he says.
That high level in the hospital definitely includes more than the CIO. At Sentara, Reese says the project management office, chaired by a vice president, runs the project and keeps executive alignment in order. Sentara also had a design committee made up of the senior executives, including the CEO, that met Thursday mornings to discuss decisions affecting the health system. “We blocked their calendars,” says Reese. Though it's not always easy to get a CEO to commit to that many meetings, he notes, “My burn rate is $4 million a month — and a CEO will walk in the room for $4 million a month.”
Ganguly had a similar setup with his executive oversight committee. “It was chaired by our CEO,” he says. “I, as the CIO, was the facilitator, but ultimately the CEO ran the committee and set the strategic direction.”
At Sharp, the executive sponsor for the project was Sharp's executive vice president of hospital operations. “We were very deliberate in making this a business decision, not an IT decision,” says Spooner.
And, according to Thompson, hospitals need to align the organization differently, too. They have to take into consideration the support of legacy applications and the implementation of the new applications. Thompson says there's a need to keep the current train running while setting up another train on a parallel track. “At some point, the engineers hop over to the new train and you're on a different path. But you've got to keep the old one running.”
So do CIOs need to restructure their IT teams to run two trains at the same time? Absolutely, says Thompson. “Because you've got two camps — you've got the legacy system to keep the doors of the hospital open, and the new structure with project managers and implementation specialists who need to be trained. He says it's important to include the legacy discussion in that planning. “You can't implement the new without understanding where you're at with the old,” says Thompson. “The question is, how do you keep that legacy staff engaged in the implementation?”
Some organizations use a dual reporting structure for a few people in that legacy vertical. Their primary responsibility is keeping the legacy applications operating and ensuring the customer's needs are being met. But they're also contributing to the new application development. “They're bringing that legacy application and workflow expertise to the table and getting their hands dirty in the build,” says Thompson.
Many say that's a big challenge, as it requires a new line of thinking for rejiggered IT staff. “There are a lot of turf wars and thinking they're going to be left behind,” he adds. How do you resolve that conflict? Thompson says it's about over-communicating and ensuring that affected staff is involved in the new discussions.
And some consulting organizations are jumping in by pitching legacy support applications while the hospital team moves to the new. “It works and it doesn't,” says Thompson. “You can't just farm it all out because of the history and the background — it's a delicate balance.” Outsourcing the legacy system, however, is still the choice of many.
At Sentara, Reese went for an outsourcing deal. “Do you leave your current staff on the old system and hire new staff to work on Epic, or bring your old staff to work on Epic and do something else with your old architecture?” he asks “We outsourced (the old) to Perot” (Plano, Texas). He says his three-year, $25 million deal has allowed him to take his current staff (and their institutional knowledge) and train them on Epic. “The staff loved it.”
Ganguly was another CIO who chose outside assistance. “Meditech did offer some assistance, but in the end we went to a consulting group (the Shams Group, Coppell, Texas) that were specialists in the Meditech arena. Of course, he adds, CentraState had to budget for that as part of the conversion.
“I think where we screw up as CIOs is when we negotiate contracts like this, we focus so much on the cost of the hardware and software and we negotiate the heck out of it — and then we nickel and dime the implementation and cut the implementation cost to almost nothing.” He says that's where the real cost is — “having enough people assigned to get this thing done.”
At Sharp, Spooner had more than 100 people involved in the project teams. “The whole group spent two weeks in Kansas City (at Cerner headquarters) — and it wasn't recreation.” That knowledge was then extended out to the 300-400 physicians who would be using the system. But Spooner says that turned out to be a real challenge. “Early on, we realized we hadn't done as much communication as we needed to have done.”
Time on our sides?
Once the staffing is aligned, what sort of timelines can a CIO expect?
Many say a three-year project is typical. Thompson says cutovers are incremental, and hospitals will be running dual systems for a while. “It’s a very delicate balance, because you’ve got feet in both worlds, and when you’re talking about clinical applications and process, there’s a lot at stake.”
He says the CNOs and CMOs will get uptight in a hurry if there isn’t a very clear plan. “The Swiss cheese holes should not light up when you’re doing a conversion, because that’s when people die.” Thompson says the biggest challenge in migrating is continuing access to the legacy data which typically is not all in one place. “How do you get that over and keep it meaningful in the new application without creating safety issues during that transition period?”
Ganguly, who went live in January 2007, had a three-year process that called for running both systems in parallel. “It’s not cut and dry because we were live in test systems first and then we did a ‘live in dark’ where we took the Siemens system live with real live interfaces and data flowing.” He says keeping things coordinated and synchronized was a challenge, but using a project team structure helped. Ganguly also created the role of director of IS, which was ultimately filled by the vendor’s project manager.
Sentara signed its contract with Epic in 2005 “Last year, we rolled out all the EDs, scheduling and two or three practices to make sure the dog would hunt,” says Reese. His timeline? 2006 was the build year, 2007 was finish the build, testing and rollout ED, and by 2008 he was getting his first hospital on full CPOE. “This is our timeline with Epic’s approval,” says Reese. “It’s hard to say whose timeline it is because you negotiate it. They give you advice, and when you both get done with it, you both think it’s your own.”
Spooner says the timeline was a give and take with Cerner. “We went with implementation using Cerner methodology.” That meant Cerner presenting Sharp with a variety of choices in system configuration and Sharp picking one that most closely matched its workflow. “Basically they said if we accepted the Accelerated Solutions method and agree to a single build for all the hospitals, they would commit to a certain implementation timetable.” He says that while some think such an approach is limiting, the structured vendor methodology gets makes implementation much faster.
Once the implementation timetable is set, Thompson says change management need to be addressed. “How do you get the IT organization — and also the operational folks, physician groups and revenue cycle people — bought in and up to speed?” Though downtime is always an issue, in conversions like this Thompson says the most pressing downtime is not a hard downtime where there is no access to data; rather, it’s an intellectual downtime where staff says, ‘Oh my gosh we’re doing it differently now.’
Thompson believes many traditional IT professionals still think conversions consist of ‘bringing data from point A to point B, turn it on and you’re good.’ “It doesn’t work like that,” he says. “Ninety percent is the organizational change, supporting the end users, getting them to be efficient and avoiding the safety issues.”
Ganguly says his starting point was documenting the physical processes of the EMR. Though the vendor gave some help, he says it is work that must be done internally. “Nobody knows how you do business better than your people,” he says. “The problem we found is, depending on which people you talk to, the way they do business is a little bit different.”
That lack of standardization hindered his ability to roll out automation because there were too many exceptions. His first step was tasking a standardization committee to look at process and terminology to make sure things like discharge date meant the same thing to every employee. “The technical stuff isn’t necessarily the more difficult part,” he says. “It’s the culture change and the process change that are huge . Moving the data is the least of it.”
Part of the culture change is getting organizational buy-in, and Ganguly used a change management committee, chaired by the vice president of human resources for communications, education and training. Its job was to educate the entire employee population as to ‘why we’re doing this.’
“We did it over and over and over again,” says Ganguly. “You had to, because it’s a long process and people lose sight of the rationale.” He says during implementation, every employee knew what the hospital was doing and why. “Whether they agreed with it is a different issue, but at least they understood.” Ganguly said this helped him tremendously because, as a community hospital struggling with resources, it’s impossible to take nurses from each unit and assign them to IS for two years. “It’s something they had to do in their spare time, and we really had to have buy-in for that.”
Reese got buy-in early by involving the organization in the selection process. “We got their opinion about what it would do to their business,” he says. Reese says that, in the end, 3,000 people participated in the review. “I needed their emotional buy in.”
Thompson explains that along with change management, access and security are areas where hospitals often fall down in a conversion. “You can’t test and validate and verify enough.” Confidentiality and HIPAA are also an issue. “I can’t tell you how many times I’ve been in the command center and you flip on the lights, you’re in the new space and 90 percent of the calls the first day are, ‘I was supposed to have access to this, and I don’t.’”
He says one school thought recommends giving generous access and evaluating it post-go live, then lowering it. “With the other theory, if you lock it down and just give people what they need, you’re going to have a ton of issues during go-live,” he says. “Giving folks access to do their jobs is critical, especially in the clinical setting.”
And though it may seem contradictory to the spirit of implementing an electronic record, printing is an issue as well. There are documents that are going to be printed, like prescriptions and discharge instructions. During go-lives, because of volume, those often go to a wrong printer or don’t print at all. “Every implementation I’ve ever been a part of, security and printing are the top two issues,” says Thompson.
His key takeaway is keeping conversions a partnership model with the new vendor and the legacy vendor, ensuring candidness in all relationships and clearly defining expectations starting from the contracting phase. “As long as you’ve got those things covered, you’re OK.”
Reese says it’s important to remember that there’s more than one way to do a conversion. “This is the way we did it,” he says. “Some people can do it differently and are very successful. It’s not one size fits all.”
And after having gone through the process, Spooner points out something many forget about EMRs: “As you become more electronic, you become less attuned to the backup systems you used to have,” he says. “We have to plan a lot more carefully for these transitions because, at some point, there may not be a paper chart to go back to.”
He also says next time around he’d do a better job in the communications department. “We improvised on the communications and needed more and more consistent communication. We hope that we’ve learned those lessons.”
Ganguly shares a similar take, and says that, in the end, it’s about education, communication and culture change. “Get those three things done, and the technical stuff will fall into place.”