When CIO John Hummel tried to computerize the way doctors write prescriptions for patients at Sutter Health, a network of 26 hospitals, he came smack up against an immovable obstacle: physician resistance.
"We tried to implement computer physician order entry four or five different ways," says Hummel. "We tried four different vendors. We tried PCs, tablets and PDAs." In every case, technology slowed the process for the physicians. A prescription that took five seconds to order on paper took two to three minutes on the system. The doctors were frustrated over the crippling effect on their productivity, Hummel says.
The result: "Even with the (improved) system we finally implemented, it takes about a year for a physician to recover to 90 percent of their original productivity," he explains, and less than 20 percent of the doctors use it. "I would have to label the attempt a resounding failure."
User resistance is nothing new in IT projects, but in a hospital setting, the stakes are higher, and life-saving projects can be doomed by recalcitrant doctors. Still, user opposition is justified if new safety systems steal precious time and disrupt workflow.
The best way to win doctors over is to follow the same rules that guide any good IT system implementation: Start with strong executive support and realistic expectations; understand your processes and how the system will affect them; anchor the system with an efficient, bug-free back end; build a user-friendly front end; find committed users to lure in the rest; and provide good technical support. But most important, understand what really motivates your users.
Failed Front Ends
Hospital safety systems are doomed to fail if the physician-facing front end isn't tied to an efficient and responsive back end, and the back end needs to be built first, says Chris Giglio, director of the Accelerated Solutions Center at Cerner, a vendor of hospital automation systems. "The physician changes should be implemented last," he says. The system must have a flawless workflow before it's put in physicians' hands so that the effect on them is minimized.
Jayashree Raman, CIO at The Reading Hospital and Medical Center, is following that order. Last year, she installed a back-end medication administration records (MAR) system, which enables clinicians to match a patient's bar code to the bar code on a prescription drug before it's administered. "The MAR is designed to ensure the five rights: right patient, right drug, right time, right dosage, right route," Raman explains.
She also plans to implement a computerized physician order entry (CPOE) system, but that will take time. "It is a complicated project affecting multiple departments," Raman says. "We are working through the process issues and anticipate having a working model in the next few months."
Tomorrow wouldn't be soon enough for Terrence Cescon, a doctor at Berks Hematology-Oncology Associates, who works with Reading. "Reading Hospital is in the Stone Age," says Cescon, who trained at Hahnemann University Hospital, where a CPOE system had already been implemented. Because he's familiar with the technology, ordering a prescription online is quicker for him than writing it by hand, he says.
But Hummel maintains that the back-end MAR systems are even more important than the front-end physician systems.
"It's popular to point the finger at doctors' horrible handwriting," he says, referring to the 1999 Institute of Medicine report that estimated that as many as 98,000 patients die each year as a result of preventable medical errors in hospitals. But more recent studies have shown that only 13 percent of the errors are due to poor handwriting, he says. Eleven percent are introduced at the pharmacy, and a full 50 percent of the errors are made at the bedside.
"You can spend millions trying to avoid a physician's handwriting, but it won't make that much difference," he says. "Hospitals can make a much bigger impact with bar coding."
Even so, MAR systems alone aren't good enough, says Claire Turner, director of communications for The Leapfrog Group, a consortium of companies that provide health care to 35 million Americans. "Implementing only the back-end process doesn't deal with the real issue," she says. "The earlier in the process you catch the errors, the better."
Turner compares a fully automated electronic health record system (which includes both CPOE and MAR) to safety systems in a car. "We need a seat belt, an airbag and antilock brakes," she says.
John Glaser has had experience at both ends. At Partners HealthCare System, a nonprofit network of 10 hospitals, he's implementing CPOE systems at several of his teaching hospitals.
"It's a bear -- very complicated, highly invasive of workflow, constantly encountering challenges," he says. But unlike Hummel, he doesn't use the term failure. "No project is a complete success or a complete failure," he says. "It's a matter of degrees. All projects have problems."
For example, Glaser tells of rocky times during a pilot of an MAR project. "The wireless PDAs and the network were flaky; they just didn't work reliably," he says. "Plus, we didn't really think through the workflow."
Workflow problems can sabotage a hospital safety system because doctors won't stand for the disruption. "If you do it right, implementing a whole system is like ripping out the backbone of a hospital and replacing it," says Paul Ruflin, CEO of Eclipsys Technologies, a provider of hospital medical record software. "It is pervasive technology."
"The challenge is redesigning the entire workflow," says Jacque Dailey, CIO at Children's Hospital of Pittsburgh. "Success is based upon process improvement, not the technology."
And process improvement requires strong support -- from both executives and IT. Dennis Baker, CIO at Sarasota Memorial Hospital, a community hospital, has been learning about that since 1999, when he implemented a CPOE system without enough support from either group. "For years, we only had 25 percent of the orders placed online," he says.
Executive backing, although belated, made all the difference. "We finally got the CEO involved," Baker says. The CEO sent a letter to all the physicians, mandating the use of the system within a year. "Now we are proceeding floor by floor with extra support for the clinicians using it," he says. "We are up to 50 percent and expect to get to 80 percent by the end of the year."
Although automated hospital systems are expensive and disruptive and doctors complain about lost productivity in the early phases, helping everyone focus on the big picture can improve the outcome. "You have to do this for the right reasons," Hummel says, "and return on investment isn't one of them."
Hummel says that motivation can make all the difference in physicians' acceptance of a new system, recalling a project designed to ensure that specially trained doctors called intensivists monitor critical-care unit patients. "We implemented a video system so that intensivists can monitor 50 beds at a time instead of 10," Hummel explains.
Because there are only a few intensivists available, they were thrilled to be able to monitor more people during each shift. "We decreased fatalities by 25 percent," he says. "We proved that it works. The cost is over US$5,000 per bed per month, but we saved lives."
Jocelyn Benes, vice president of quality control at Children's Hospital of Pittsburgh, says that when the hospital implemented a CPOE system, "we didn't even measure financial benefits. Our focus on quality was one of the reasons we were successful."
Benes says that appealing to users' dedication to the job is the key to getting their buy-in. "Clinicians are more receptive to doing what's best for the patient than they are to saving money," she explains. "We were successful in decreasing errors, and that's all that mattered to the patients and the staff."
Different Players, Same Game
CIOs and specialists who have implemented hospital safety systems say that while the players are different and the stakes are higher than for other IT projects, most of the success factors are similar.
Drive the project from the operational area of the hospital, not IT.
- Use a multidisciplinary team that includes doctors and nurses to design the process.
- Get buy-in from all stakeholders, especially top management and the top doctor in each department. Forcing a system on clinicians who don't see the benefits is a prescription for failure.
- Keep all stakeholders informed throughout the process. Clinicians often don't have offices and don't check e-mail, so devise special methods of keeping them up to date.
- Enlist the nurses and other practitioners early and use their feedback to improve the process before attempting to convert the physicians.
- Provide good support and resolve problems quickly. Have people on-site during the rollout phases -- especially when physicians are involved.
- Make sure the system is ultrareliable. Keep antivirus software up to date and fully test all patches before implementing them. A computer virus isn't normally a life-or-death matter -- unless the computer is monitoring someone's life support.
- Set the proper expectations for the project. Don't try to do too much too soon, and explain about trade-offs, such as time versus access to information.
- Focus on measurable quality and safety goals that relate to saving lives. Don't focus on financials or expect the system to save money.