The Trouble with EHRs

Although electronic health records are known to reduce healthcare costs, concerns about accuracy and usability—and the risk of EHR-caused medical errors—are growing.

This article appears in the December issue of HealthLeaders magazine.

The move away from paper medical records to EHRs has many benefits, but the flip side is that providers need to carefully manage the usability, accuracy, and audit trails of EHRs across the entire care team.

Providers say the day of the doctor carefully controlling and supervising every aspect of a medical record is giving way to an age where the entire medical team and even patients will scrutinize and update a record's accuracy, all while legal and regulatory requirements for maintaining those records are met. At the same time, the complexity of EHR software itself increases the risk of EHR-caused medical errors.

While EHRs have been around for years, the requirements of meaningful use stage 2 for patients to be able to view, download, and transmit their medical records to providers across the care continuum as well as health information exchanges mean it is time for providers to pay special attention to proper record maintenance.

Dealing with errors in the record is coming to the forefront of these concerns. "Traditional practices within the paper record support a single-line strikethrough of the original documentation," according to Amendments in the Electronic Health Record Toolkit, a 2012 paper published by the American Health Information Management Association. "However, these practices will not necessarily transfer to an electronic environment, and new practices should be evaluated against organizational policy and specific system limitations."

AHIMA's paper also advises: "Processes for receiving the patient amendment request identifying PHI affected, determining whether it should be accepted or denied based on the type of request, and notifying the patient of the outcome must be developed."

Further complicating the picture: As physicians move from clinic to clinic or hospital to hospital during their workday, they are using EHRs from multiple vendors, and each EHR may express discrepancies in the medical record differently.

"They may have multiple different interfaces to deal with," says Blackford Middleton, MD, chief informatics officer and professor of biomedical informatics at Vanderbilt University Medical Center, a 626-licensed-bed hospital in Nashville.

Although EHRs improve quality and patient safety and are known to reduce costs, there may be problems associated with their use, and they can even cause medical error, he says.

Middleton likens the abilities of today's EHRs with those of the challenges of using word processing in the late 1980s, when such software ran into usability and interoperability problems.

"Standardizing interoperability means not only the idea of sharing data between system A and system B, but also the idea that the clinician who's rounding at the hospital in the morning, seeing patients all day in [the] clinic, and rounding at a different hospital in the afternoon, can have a calming sort of user experience, if you will," Middleton says.

At the same time, he adds, "We're moving well away from kind of the doctor-on-a-pedestal or omniscient, all-knowing clinician, to someone who, ideally, is collaborating with a patient, designing care goals together, and helping the patient toward those goals in a very effective manner."

Middleton arrived at Vanderbilt in February 2013, having previously been at Boston-based Partners HealthCare and Harvard Medical School. Vanderbilt uses a homegrown EHR, and over time the medical center is showing more of this EHR to patients. "There are reports from operating rooms and procedures and whatnot, but we don't show the full notes yet," he says.

"With docs and nurses and even patients now documenting into electronic tools, what's the quality of the information as documented? For the legal record to evolve to be inclusive of the electronic record and its core components, we have to have the same kind of amending or addending process that is the rule, or the standard of practice, for the paper-based record," Middleton says.

In 2012, Middleton coauthored a paper for the American Medical Informatics Association, Enhancing patient safety and quality of care by improving the usability of electronic health record systems: recommendations from AMIA. The paper recommends that providers give systematic feedback to EHR vendors to improve EHR usability and safety.

Meanwhile, patients are more and more able to suggest annotations for the clinician's record. The Open Notes movement—which began as a Robert Wood Johnson Foundation study at Geisinger Health System and research done at Beth Israel Deaconess Medical Center and Harborview Medical Center—has spread to the Cleveland Clinic. All told, including patients of the Veterans Administration, 1.8 million patients nationwide are able to access their medical records.

In the event of patient feedback, "it's up to the clinician to write something useful in the record to take note of the patient's input," Middleton says. "There has to be an auditable trail of all such changes to the record, as the record becomes increasingly an important point of collaboration and communication between doctor and patient, and nothing should ever be deleted. It may be suppressed and invisible, but one needs to be able to review that through an audit kind of report."

All these changes reflect a groundswell in modern U.S. healthcare toward much more transparency in everything providers do, Middleton says. "There's no point in resisting it. We need to be transparent in the services we deliver, in the cost of care we deliver, and the value, so that employer purchasers, patients, and everybody can make informed choices. Make sure you write your notes with the idea in mind that they will be public."

For that matter, consider that maintaining the quality of an electronic health record is an ongoing responsibility of the entire medical team, says Rachel Chebeleu, corporate director of professional fee abstraction at the 805-licensed-bed Hospital of the University of Pennsylvania in Philadelphia.

"You need people who are concurrently, proactively auditing your records for copy-and-paste issues, for copy-forward issues, and for just general compliance," Chebeleu says. Those tasks should not be relegated to billing specialists or coders but instead be assigned to staff who will audit records for their clinical quality, she adds.

"There might be some nurses who really know clinical content well, are not interested in being on the units anymore, and will just go through records and audit them, the important record-type entry stuff," Chebeleu says. The same audits can examine practice patterns, she says.

EHR audit productivity also needs to be overseen. "You don't want someone who gets so immersed in the record they are able to get through only two a day," Chebeleu says with a laugh. Penn Medicine, the system to which the Hospital of the University of Pennsylvania belongs, uses Allscripts' Sunrise Acute Care (formerly Sunrise Clinical Manager) as its EHR for inpatient computerized physician order entry and discharge summaries, and is now implementing physician notes starting with its rehabilitation facilities.

In its outpatient practices, Penn Medicine uses Epic EHR software, which in the next few years will replace Allscripts on the inpatient side, as well, Chebeleu says.

The governance challenge of achieving quality in EHRs is foremost in Chebeleu's thoughts. "Get the people who care about it" involved, she says. "Get risk management on board, your general counsel on board, your compliance department, your coders, even the quality people," she says. Find the necessary funding and tackle the problem proactively, she suggests.

"This will cut off at the pass a lot of problems you could have if you don't do this," Chebeleu says. "Even if you invest one full-time equivalent, pilot it."

The move to a team-based effort to improve the quality of the medical record "flies a lot in the face of our traditional conception of the medical record as something that belongs to the doctor [and] belongs [only] to the health system," says Jeffrey Linder, MD, a general internist in the division of general medicine and primary care at the 800-bed Brigham and Women's Hospital, part of the seven-hospital Partners HealthCare system.

Until recently, Lindner served on the executive committee of the Longitudinal Medical Record, Brigham and Women's homegrown electronic medical record. Linder still serves on the adult primary care expert panel at the hospital and, like almost all Brigham physicians, has a faculty appointment at Harvard Medical School as an associate professor of medicine.

"It's a big shift to move from that to a more collaborative space where patients can enter data that's useful for their care, that their doctors want to see, but it's also curated and accepted into the record by some clinician," Linder says.

The big flood of nonphysician, nonclinician–entered data that Linder has anticipated for the past five years has yet to arrive, but he fully expects it.

Toward that end, Partners is creating a section in its EHR for patient-entered data. A clinician will need to review the entries and accept them into the chart, but those entries will remain labeled as patient-entered data. Data that is not accepted into the record will also remain, but "my impression is you'd really have to go looking for stuff that wasn't accepted into the record," Linder says.

At the same time that this is planned, Partners is in transition from LMR to Epic, due to be deployed across both inpatient and outpatient systems in 2015, Linder says. "The new healthcare landscape will challenge us to engage in population health management, improve the coordination of healthcare, and accept financial risk for the care of our patients. This new system will enable us to meet those challenges," Linder says. Between now and then, Partners will conduct due diligence to make sure that Epic's capabilities allow the kind of audit trail and other features it requires, he adds.

"We're very interested in having patients enter their own blood sugars and blood pressures and having some mechanism of allowing them to," Linder says. "The only mechanism right now for them to correct their data is via messaging. If they notice a problem that they disagree with, currently it's not really a collaborative space, where the practice and the patient are jointly curating the information in the medical record."

Like the other providers, Partners faces its own governance challenges during this transition. "The other cultural problem we're running up against that we're going to have to work our way through over the next decade is our conception of medical care—the unit of medical care," Linder says. "Right now I work in the ambulatory setting, so we're very focused on visits still, and hospitals are very focused on hospitalization, and so the record is designed around discrete events like that."

Thus, documentation of a medical record today remains largely the story of what was done at a particular visit. "Hopefully we're moving toward a health system that's focused on taking good care of you, whether or not you happen to be in the office at that moment," Linder says. "That shifts the value from visit-based data to the latest instance of the electronic health record as the truth, and that can happen at any time. It's not dependent on a visit."

Of all the stakeholders in healthcare, the ones most concerned about how the medical record evolves are probably the health systems' attorneys. "It's our lawyers who are most interested about keeping a record of everything that happens," Linder says. "Our health information systems department is very focused on keeping a record this way."

Even now, LMR records have a "history" button next to many coded elements, Linder says. "You can kind of delve into the history and see who updated it, and how and when," he says. "But if you think about all the moving pieces in an electronic health record, every single one of those elements has some mechanism for capturing histories behind that element, and it's a lot of data that nobody really has the time or inclination to go through, so it's challenging. It's definitely challenging."