Why Electronic Health Records aren't more usable
Physicians often have difficulty entering structured data in EHRs, especially during patient encounters. The records are hard to read because they're full of irrelevant boilerplates generated by the software and lack individualized information about the patient.
Alerts frequently fire for inconsequential reasons, leading to alert fatigue. EHRs from different vendors are not interoperable with each other, making it impossible to exchange information without expensive interfaces or the use of secure messaging systems.
EHRs are designed to support billing more than patient care, experts say. They add to, rather than reduce, the workload of doctors. And they don't follow the principles of user-centered design (UCD), which puts the needs of the user at the forefront of the design and development of products and systems.
[Related: Health IT glossary]
The American Medical Association in 2014 issued an eight-point framework for improving EHR usability. According to this framework, EHRs should:
Nevertheless, it does not appear that EHR vendors are placing more emphasis on UCD. The Office of the National Coordinator for Health IT requires developers to perform usability tests as part of a certification process that makes their EHRs eligible for the government's EHR incentive program. Yet a recent study found that, of 41 EHR vendors that released public reports, fewer than half used an industry-standard UCD process. Only nine developers tested their products with at least 15 participants who had clinical backgrounds, such as physicians.
Government regulations are a key barrier to improving the usability of EHRs, says Peter Basch, MD, medical director for ambulatory EHR and health IT policy at MedStar Health, a Washington, D.C., healthcare system. For example, he notes, EHRs were easier to use before developers were required to design them in a way that enables them to collect quality data for the government's EHR incentive program. (Physician practices and hospitals must gather this data on health care processes for a certain percentage of Medicare or Medicaid patients to show "meaningful use" of EHRs so they can qualify for incentives and/or avoid financial penalties.)
Another downside of the meaningful use program, says Basch, who advises the American College of Physicians on health IT, is that vendors have had to focus on rewriting their software to meet the changing EHR certification criteria. As a result, they have had little bandwidth left over to meet the needs of their customers by building more user-centered products.
The other big regulatory obstacle is the billing guidelines of the Centers for Medicare and Medicaid Services (CMS), which are followed by most private insurers. The guidelines for "evaluation & management" (E&M) billing codes specify that physicians must document that they have performed a certain number of services to claim a particular coding level for the length and complexity of patient visits.
Because the amounts that doctors can charge for visits depend on these codes, EHR developers designed their products to help physicians justify their coding levels in case of an audit. This was a big selling point, especially in the early years of EHRs.
However, this approach has had some unintended consequences. First, it forces physicians to follow EHR templates of drop-down click boxes that do not necessarily reflect how they conduct encounters with patients. Second, it makes them spend too much time entering data. And third, the resulting computer-generated notes are often unreadable.
Part of the standard templates that doctors are supposed to follow, for example, is the "review of systems," which covers all of the body's physical systems. In an EHR, the review of systems may include hundreds of check boxes. Most of these are inapplicable if, for instance, a physician is setting a broken bone, notes Mark Anderson, a health IT consultant based in Montgomery, Texas.
Some EHR vendors tell physicians to simply use a macro that checks off all of the boxes as "normal" findings, he says, and then change the ones that are not normal. But when the EHR converts the structured data into text, this approach generates five pages of descriptions of the normal findings, which are irrelevant to the case.
Basch says the E&M coding guidelines must be reformed before EHRs can become truly usable. But that's unlikely to happen as long as fee for service remains the predominant method of physician reimbursement. The meaningful use program, meanwhile, has no fixed endpoint; in fact, the third stage of the program is scheduled to begin in 2017.
Doctors' biggest complaint about the EHR is that it slows them down, especially in the documentation phase. "Compared to handwriting or dictating, EHRs take doctors nine times longer to enter the data," Anderson says. "Sure, you have more information in the EHR than in paper records, but it takes more time."
Eventually, he says, natural language processing (NLP) will become good enough so that it will be able to extract most relevant concepts from dictated text and place them in structured fields. Considering that medications, diagnoses, and lab results are already coded, he notes, NLP engines have to convert only a relatively small portion of text into structured data.
Meanwhile, however, many physicians stumble along, working longer hours to get their visit notes into the EHR. Doctors who sluffed off the initial training are in much worse shape than those who paid attention and tried hard to learn the system, Basch points out. Mature users like himself can document fairly rapidly by using a combination of point and click, typing, and dictation with speech recognition. But physicians who are fairly new to EHRs—a majority of users at this point—may have a lot of trouble keeping up with the flow of information.
Many doctors have developed workarounds. Either they dictate the majority of their notes, which produces a mass of unstructured text, or they copy and paste parts of previous notes into current ones. The latter method not only increases the risk of errors, but also may create the appearance of fraud. And it increases the problem of locating important information later on in the copious, repetitious text while trying to treat patients.
Beyond documentation, EHRs generally include alerts that are supposed to improve quality and safety. These include alerts that pop up when a patient needs recommended preventive or chronic care. Some of those reminders are irrelevant, either because a patient doesn't fit the parameters or because the care was performed elsewhere.
Other alerts go off to prevent adverse drug interactions with other medications, allergies, or foods. Many of these are inapplicable to particular patients, and after a while, doctors may stop paying attention to them or turn them off. Three quarters of EHRs don't allow the customization of these alerts, according to Anderson.
[Related: Health IT use lags despite consumer interest]
Most EHRs were designed for primary care, so they don't work well for other kinds of specialists, he notes. Some good specialty EHRs exist, but not in all specialties, and those EHRs may not be interfaced with an organization's enterprise system. Epic and Cerner, which make the leading EHRs that cover both hospital and ambulatory care, provide decent templates for inpatient specialists, Anderson says.
To save time, Basch would like to have an EHR that can reuse information. That would enable him, for example, to use the data already in the system to populate documents such as prior authorization forms.
He'd also like a "smart" EHR that places information in context. "When I look at a lab result for liver function, I don't just want to see prior results, I want to look at other things if they're elevated," he says. "Show me a med list and show me the meds that the patient is on that could possibly impact liver function. Or show me imaging studies. Because right now I do that manually."
Anderson hasn't seen any EHRs that can reuse data or apply it intelligently, as Basch describes it. This is a next-generation concept for vendors, he says. In fact, he hasn't even seen an EHR with alerts that correlate medications with lab results.
What many physicians want in an EHR, Basch says, is something as simple and intuitive to use as an iPhone, but he thinks it's a mistake to "dumb down" these systems.
"That would cause much of what I'm expecting from my EHR to disappear. In fact, I want it to help me with the complexity of patient care. I want it to present me with a rich context as simply as possible in a way that I can see it and not ignore it."