The incident that perhaps most fully impressed the potential dangers of electronic health records (EHRs) on hospitalist pioneer Robert Wachter, MD, MHM, came two years ago. It started, innocently and well-intentioned enough, years earlier with the installation of EHR systems at the University of California at San Francisco (UCSF). Flash-forward to 2013 and a 16-year-old boy’s admission to UCSF’s Benioff Children’s Hospital for a routine colonoscopy related to his NEMO deficiency syndrome, a rare genetic disease that affects the bowels. For his nightly medications that evening, the boy was supposed to take a single dose of Septra, a common antibiotic that hospitalists and internists across the nation routinely prescribe for urinary and skin infections.
But this boy took 38.5 doses, one pill at a time.
How could that possibly happen? Hospitalists might rightly ask.
Because the EHR told everyone involved that’s what the dose should be. So every physician, pharmacist, and nurse involved in the boy’s treatment carried out the order to a T, discovering the error only when the teenager later complained of anxiety, mild confusion, and tingling so acute he felt “numb all over.”
In an era when EHR is king, an adverse event such as a 39-fold overdose is just another example of the unintended consequences technology has foisted upon hospitalists and other providers in America’s massive healthcare system. It is the unfortunate underbelly of healthcare’s rapid-fire introduction to EHR, thanks to a flood of federal funding over the past 10 years, Dr. Wachter says.
“Most fields that go digital do so over the course of 10 or 20 years in a very organic way, with the early adopters, the rank and file, and then the laggards,” Dr. Wachter said at SHM’s 2015 annual meeting in Washington, D.C., where he recounted the UCSF overdose in a keynote address. “In that kind of organic adoption curve, you see problems arise, and people begin to deal with them and understand them and mitigate them. What the federal intervention did was essentially turbocharge the digitization of healthcare.”
And with the relative speed of digitization comes unintended consequences, including:
- Unfriendly user interfaces that stymie and frustrate physicians accustomed to comparatively intuitive smartphones and tablets;
- Limited applicability of EHRs to quality improvement (QI) projects, as the systems are, in essence, first constructed as billing and coding constructs;
- Alert fatigue tied to EHRs and such medical devices as ventilators, blood pressure monitors, and electrocardiograms desensitize physicians to true concerns; and
- The “cut-and-paste” phenomenon of transferring daily notes or other orders that’s only growing as EHRs become more ubiquitous (see “CTRL-C + CTRL-V = DANGER”).
“Health IT [HIT] is not the panacea that many have touted it as, and it’s really a question of a reassessment of where exactly we are right now compared with where we thought we would be,” says Kendall Rogers, MD, CPE, SFHM, chief of the division of hospital medicine at the University of New Mexico Health Sciences Center in Albuquerque and chair of SHM’s Information Technology Committee. “I think our endpoint—that we’re going to get to—this is all going to result in better care. But we’re in that middle period of extreme danger right now where we could actually be doing harm to our patients but certainly are frustrating our providers.”
HIT’s rapid evolution starts with the creation of the Office of the National Coordinator for Health Information Technology (ONC) in 2004, which began receiving funding in 2009 to the tune of $30 billion to improve health information exchanges between physicians and institutions.
The money “spent in adoption should have been spent in innovation and development and research to show what works and what doesn’t well before you started pushing adoption,” Dr. Rogers says. “But at this stage, we can’t go backward … the plan is in flight, and we have to try to repair it in the air at this point.”
To that end, The Joint Commission in March 2015 issued a Sentinel Event Alert to highlight that the safest use of HIT still needs structural improvement. The Joint Commission analyzed 120 sentinel events (which it defines as unexpected occurrences involving death or serious physical or psychological injury or the risk thereof) that were HIT-related between Jan. 1, 2010, and June 30, 2013. Eighty percent were issues with human-computer interface, workflow and communication, or design or data issues tied to clinical content or decision support.
“As health IT adoption spreads and becomes a critical component of organizational infrastructure, the potential for health IT-related harm will likely increase unless risk-reducing measures are put into place,” the alert stated.
To that end, The Joint Commission recommends:
- Focusing on creating and maintaining a safety culture;
- Developing a proactive approach to process improvement that includes assessing patient safety risk; and
- Enlisting physicians and administrators from multiple disciplines to oversee HIT planning, implementation, and evaluation.
Terry Edwards, chief executive officer of PerfectServe, a Knoxville, Tenn., firm that works on healthcare communications systems, says that a survey his firm conducted in 2015 found that, among clinicians needing to communicate with an in-house colleague about “complex or in-depth information,” an EHR is used 12% of the time. Just 8% of hospitalists surveyed used it. The rest used workarounds, face-to-face conversations, and myriad customized solutions to communicate.
“Workarounds happen all the time in healthcare because many of the tools and technologies impede rather than enhance a clinician’s efficiency,” Edwards says in an email to The Hospitalist. “It’s pretty clear that many physicians are frustrated by EHR technology.”
The natural question around unintended consequences: Why didn’t physicians or others see them coming as EHRs and HIT were burgeoning the past decade? Dr. Rogers says that hospitalists and physicians weren’t involved enough on the front-end design of EHRs.
So instead of systems that have been built to be intuitive to the real-time workflow of hospitalists, nurse practitioners, and physician assistants, the systems are built more for back-office administrative functions, he adds.
“When we have programmers and non-clinical people trying to build products for us, they’re dictating our workflow,” Dr. Rogers says. “In many cases, they don’t understand our workflow, and in many more cases, our workflow differs from the last person or the last hospital they worked at.
“This is where we get into issues around usability.”
Take the overdose patient at UCSF. One wrong number typed into a single field led to the oversize dosage. Safety redundancies built in the system flagged the excessive dosage each time, but at each point, a human decided to keep the dosage at the incorrect size because, essentially, everyone trusted the EHR.
All of those red flags come with their own unintended consequence: alert fatigue.
“When people really get fatigued with all of these alerts, they start to ignore them,” says hospitalist Cheng-Kai Kao, MD, medical director of informatics at the University of Chicago Medicine. “So now here comes the question: How do we properly set the limit or threshold?”
In the airline industry, alerts are often tiered to give pilots an immediate sense of their importance. But Dr. Kao says the typical EHR interface is not that advanced, an often frustrating trait to younger physicians accustomed to user-friendly iPhones and web applications. The same frustration often is found with the litany of medical devices hospitalists interact with each day.
“Everything is a fundamental question: How do we set up an optimal environment for humans to interact with computers?” Dr. Kao adds. “We are talking about usability. We are talking about optimizing the IT system that blends into people’s daily workflow so they don’t feel disrupted and have to develop a workaround.”
One EHR critic suggests that the proliferation of workarounds could be solved by a moratorium on further implementation and rollout of EHR systems.
“During that moratorium, there needs to be a complete rethinking of roles, i.e., who does what with these systems, and what needs to be severely rethought are the roles of who gets to do what, including data entry,” says Scot Silverstein, a health IT consultant in Philadelphia. “There’s just no way you can make entry of information into complex computer systems rapid with multiple computer screens that have to be navigated through ad infinitum. There’s just no way you can make that anywhere near as efficient, and you can’t make it less distracting and untiring compared to paper.
“I’m advocating not a return to paper but a consideration of where a paper intermediary—such as specialized forms—between clinicians and information system are appropriate.”
Silverstein says that the relatively rushed overlay of computer systems on medicine meant that corporate computing models were simply pushed into healthcare, a world that operates very differently than most other industries. He says that is why adverse events will continue to occur; why The Joint Commission felt the need to issue an alert; and why the ECRI Institute, a quasi-Consumer Reports organization for healthcare, listed “data integrity failures with health information technology systems” atop its Top 10 Patient Safety Concerns for 2014. Other EHR concerns have been on the list the past several years as well.
“The business computing model, which dates back to the days of card-punch tabulators that IBM developed in the 1920s and ’30s, really has a completely wrong model of medicine,” Silverstein adds. “Medicine is not a predictable, controlled, regular environment. It is an environment of emergencies, irregular events, unpredictability, poor boundaries. Every possible thing in the world can and does go wrong.”
Dr. Rogers agrees that HIT is not optimal, but he sees little point in a moratorium or trying to stop whatever positive progress has already occurred.
“The train has left,” he says. The best approach now is twofold.
First, Dr. Rogers urges hospitalists to formalize their HIT duties by seeing if they would qualify to take the exam for board certification in medical informatics, which was created in 2013 by the American Board of Medical Specialties (ABMS). The more hospitalists who are recognized for the work they already do with EHRs, the more they can then use their positions to help lobby their institutions for changes.
Second, Dr. Rogers wants hospitalists to work as much as possible with vendors, other clinical informaticians, and related stakeholders to help improve the existing system as much as possible. In particular, improvements could help EHRs integrate clinical decision support better, which could then serve as the foundation for research and quality improvement.
Dr. Rogers uses VTE prophylaxis as an example. Before digitalization, “we were able to build all those flow diagrams onto a sheet of paper that would have logical branching points.” Now, pull-down menus and long, one-dimensional order sets regiment what can be input, and medical logic is not the primary concern.
Often, EHR providers will say issues are tied to a lack of training.
“When a vendor repeatedly says this is a training issue, I guarantee that there is a design issue that can be improved,” Dr. Rogers says.
Instead, he and others urge third-party vendors be allowed to design programs and software that can help. He likens it to independent application developers building programs for iPhones and Androids, as opposed to firms like Apple saying that only their internally developed applications would be used.
“Apple would be nowhere right now” had they done that, Dr. Rogers says. “What made them successful was creating a marketplace that all of these individuals out there—thousands of people—could start designing innovations and applications that would fit what that population needed, no matter how small that population was.”
He says a single system, applicable across all healthcare settings, would make an “even playing field for third-party vendors.”
“I think we could get there much faster,” he says. “Within a five-year period of time, I think we could solve a lot of these issues that we’re having right now.” TH
Richard Quinn is a freelance writer in New Jersey.