July 2015

The Human Touch
By Elizabeth S. Roop
For The Record
Vol. 27 No. 7 P. 20

A Bentley University study chronicles how quality assurance measures make a huge difference in the accuracy of medical records.

Physicians armed only with speech recognition technology make critical dictation errors at a much higher rate than do experienced medical transcriptionists (MTs). Without a quality assurance (QA) function, those errors are more likely to slip through and cause harm.

Those were the findings of a 2014 study in the International Journal of Healthcare Quality Assurance that sought to determine the instances of errors made in physician dictation of medical records. Researchers found, on average, 315,000 errors in every 1 million dictations—a convincing argument for the need to employ QA processes to preserve the data integrity of patient records.

There were few surprises in the findings, according to lead investigator Gary C. David, PhD, department chair and an associate professor of sociology and information design and corporate communication at Bentley University. "It told us what we already know, which is that doctors make mistakes when they dictate. … After you've been around a while and talked to people in the transcription industry, [you know] this is not news," says David, adding that the current study is an extension of medical transcription research that's been under way for seven years. "What was interesting was the hesitancy of journals to publish" the findings.

"It's like the elephant in the room; the dirty little secret. Doctors make mistakes with dictation. Once you admit they make mistakes, it brings the conversation around to what you're doing about it," he says. "That's what people don't want to talk about. … The fact that even in current EHR systems, there's no check on [dictation] quality."

Key Findings
In "Error Rates in Physician Dictation: Quality Assurance and Medical Record Production," David and his fellow researchers worked with 79 expert MTs to identify the frequency and types of errors in 2,391 dictation files during one standard workday. It was done with the cooperation of medical transcription service organizations and professional organizations.

Among the key findings was that doctors make significant dictation errors, which for inpatient records led to 153 critical errors and 403 major errors with speech editing/recognition compared with just 20 and 82, respectively, with transcription. The most common critical errors were incorrect patient and drug names, and wrong dosages. The most common major errors were made-up words or acronyms, and gender and age mismatches.

Although it was impossible to gauge the severity and impact of any errors outside the context of the medical record's use, the researchers noted that any lack of harm should not negate the fact that the errors resulted in an inaccurate record. Further, while an error may not impact care, it still may affect other areas such as coding and billing.

Researchers noted, "While the efficiencies gained (in terms of turn-around time) through physician direct entry might be significant, these gains have the potential to be offset by errors in the record, which might give rise to numerous other problems. … Thus the 'once and done' design philosophy of EHR/EMR tools overlooks the quality assurance role of medical transcriptionists (amongst others such as clinical documentation specialists). Removing a QA step in the workflow thus can have important repercussions on documentation quality."

In other words, health care organizations that rely too heavily on speech recognition technology for dictation without proper QA by qualified MTs may be taking too great a risk with patient safety to justify any cost savings. "When you have an approach that excepts out any kind of proofreading and QA, you've got a problem," David says. "These are medical records. You're asking me as a prospective patient if my medical record is only 98% accurate, which 2% do I want to be inaccurate? We want to err on the side of caution. I understand the turnaround time and cost issues. But are you going to be penny wise and pound foolish?"

Wrong Is Normal
Susan Dooley, MHA, CMT, AHDI-F, current president of the Association for Healthcare Documentation Integrity (AHDI) National Leadership Board, shares David's lack of surprise at the findings. "The number of speech recognition errors corrected by health care documentation specialists in the study is particularly telling, isn't it?" she says, noting that a rapidly increasing volume of medical record production in the acute care setting is now being done through back-end speech recognition. "The human review of these reports provides a vital quality assurance function, and when that human is a health care documentation specialist, especially one who is certified, organizations know that deep medical and language knowledge is being used to review and edit the records to accurately reflect the clinician's intention."

Dooley points to the ECRI Institute's "2015 Top 10 Health Technology Hazards" report, which ranks data integrity in the No. 2 position, as evidence of the potential risks of inaccurate medical records. (AHDI is a stakeholder in the ECRI's Partnership for Promoting Health IT Patient Safety.)

Support also can be found in the number of EHR-related malpractice lawsuits, which doubled between 2013 and 2014. According to an article in Politico, the suits focus on mistakes such as typographical errors, speech recognition errors where key words are omitted, clinicians' use of old or inaccurate patient records when making care decisions, and nurses misinterpreting drop-down menus in EHRs and inserting misinformation into the medical record as a result.

Finally, Dooley says the numbers simply don't add up. In an informal evaluation of the economics behind reviewing back-end speech recognition, she found the relative cost of paying physicians to assume this responsibility to be $70 per hour, while a health care documentation specialist would earn $16 per hour.

Dooley also cites physician satisfaction levels, which were studied in a 2013 Rand report commissioned by the American Medical Association. Lead investigator Mark Friedberg, MD, participated in a webinar for AHDI members during which Dooley asked him to comment on the true cost of requiring physicians to ensure documentation quality. He noted that because transcription salaries were eliminated and physicians did not receive additional compensation for reviewing documentation, costs may decrease initially. However, those savings are quickly eclipsed by other expenses such as physician dissatisfaction with EHRs and increased physician burnout.

"The overwhelming lesson is that patients can be endangered when penny-wise, pound-foolish decisions like eliminating all health care documentation specialists are made in the interest of cost savings," says Dooley, adding that increased pressure from The Joint Commission will grab the attention of health care organizations that have been ignoring the issue.

"Because health care documentation quality wasn't reviewed on accreditation surveys, it was easier for administrators to ignore its necessity," she says. "But The Joint Commission is noticing health care record quality now, with plans to institute survey recommendations soon. And as you can see, increasing numbers of malpractice suits are using poor documentation quality to make a case for poor patient care. At AHDI, we're seeing increased demand for information on how to implement QA programs on clinician-created documentation, and we expect that need to continue to rise."

Striking a Balance
Despite the increase in errors from an overreliance on speech recognition technology, most in the transcription industry agree that the technology can reduce costs and streamline efficiencies when properly designed and implemented. Success lies in striking the right balance between technology tools and proper QA management.

"You can't stop technology and progress. That's not the point," says Maria French, president and CEO of clinical documentation solution provider Terra Nova. "We need to use these technologies and tools to do a better job. You don't want to [bring] it in with a sweep of the broom … Clinical documentation is at the core of every health care encounter. It must be complete, accurate, and reflect the full scope of care provided. The documentation and the data extracted from that documentation are how providers are being measured and adjudicated. Ensuring consistency in clinical documentation that's accurate, specific, and timely is a key quality measure for any facility. That's a challenge for many organizations right now.

"There's a balance out there," French continues, noting that transcription has become unfairly commoditized. "It's a huge obligation for someone to be responsible for a diagnosis, for getting the correct medications and dosages. It's a huge responsibility for practitioners. When a transcriptionist takes a document and says, 'There, it's been transcribed accurately,' and uploads it, they're affecting someone's health and sometimes their lives."

She says buying into the hype that speech recognition technology can eliminate transcription and save hundreds of thousands of dollars is a risky strategy. When done incorrectly, the resulting errors spread quickly to downstream systems where it becomes more complex and costly to correct.

In many cases, when hospitals realize their mistake, they throw more money at the problem by pushing clinical documentation improvement downstream, sometimes hiring scribes to shadow physicians. "So you're fixing what you did by removing the transcription piece [by] adding another warm body to insert the physician's dictation into the EMR/EHR," French says. "All of these steps come back to 'We took the [MTs] out of it. We don't know how to get back to where we were before, with the quality we had, so we're going to invest even more money in it.' Sometimes, no one thinks it through to begin with. … We can have a balance, but people have to evaluate their specific facility needs, and do the math."

That means including health information and transcription management in the decision-making process. Ensuring medical record and patient data experts have a voice in the move to speech recognition will help create a balanced, blended approach that leverages speech recognition to streamline processes and human intervention to maintain data integrity.

"The pendulum is swinging back and forth," French says. "We've gone from full labor service to full technology. I hope we'll see that pendulum swinging to a balance where hospitals are saying we have to get this right. ... We should have the physician dictate. It should go to draft. And transcriptionists/editors should review that document before it's complete [and] you'll have a quality document going through 98% to 99% of the time as opposed to 95% or 90%."

David, who is among a handful of researchers contributing to the scientific body of knowledge on the impact of dictation and transcription on data integrity, concurs that technology can be a valuable tool if deployed thoughtfully. A first step is to realize that transcription plays an important role in both clinical outcomes and the revenue cycle. The latter is particularly important when it comes to justifying an investment in QA as part of a technology-enabled transcription process.

"If documents generate revenue, that makes transcription and quality assurance directly tied into the revenue cycle, so you should take an interest," David says. "[Transcription impacts] not only clinical indicators and quality of care, but also return on investment, case mix, etc. It's all tied together. If you don't know that, then I don't know what business you're in."

Learning the Hard Way
David notes that dictation and transcription errors are "screams for a more intensive case study" to instill greater confidence in the message being sent to health care organizations about the need to retain and even expand the QA function. More data may drive home the message that speech recognition without QA is an unnecessary safety and financial risk.

"Managers love new technology if they think it will make their process faster. It's the shiny key syndrome. Health information technology is a shiny key. They want it because they think they should have it without thinking about the larger implementation of what they need," David says. "Everyone is so busy keeping up that no one wants to take a timeout, slow down for a second, and think about these larger issues. This is an information management issue. [Eliminating QA] works against everything we know about effective management, especially when you have errors in your process and you want to fix them by removing the quality assurance piece.

"I wish more organizations would take an interest in this. If they would, I'd be happy to talk to them."

— Elizabeth S. Roop is a Tampa, Florida-based freelance writer specializing in health care and HIT.

AHDI RESOURCES COMBAT ERRORS
To get a taste of just how serious—or in some cases humorous—speech recognition errors can be, look no further than the Association for Healthcare Documentation Integrity's (AHDI) "SR Errors—Funny or Fatal?" Facebook page. For example, the technology tool generated the following report: "X-rays of his foot were unremarkable, although they were read as having a fifth metatarsal head fracture. This was not seen on his previous CT and it is likely an ovary."

In another case, a report read: "The patient is a Seventh-day Adventist and finds it helpful to believe in cheeses."

Although many examples are amusing, some drive home the seriousness of dictation errors. In one case, the physician's instructions for a patient who was being transferred to the hospital were, "The patient has a severe allergy to penicillin and morphine and also has allergy to latex. Please use caution." The report generated by the speech recognition tool read: "The patient has to be started on urgently penicillin and give her some of the morphine. Also has elevated enzymes. Please use Anaprox."

To help combat dictation errors, the AHDI teamed up with AHIMA to create a free resource kit on instituting quality assurance (QA) programs on clinician-created documentation, which includes examples that help identify critical vs noncritical errors.

The organizations also collaborated on a best practices white paper focused on the emerging profession of clinical documentation improvement (CDI) specialists, whose function differs from that of a health care documentation specialist. Susan Dooley, MHA, CMT, AHDI-F, current president of the AHDI National Leadership Board, says, "both functions are needed" in today's health care environment.

QA programs are related to documentation integrity, which is separate from CDI efforts. A QA program for clinician-created documentation includes reviewing and editing with a focus on identifying critical errors that could compromise patient care and safety. In contrast, CDI professionals review documentation for "any clinical conditions or procedures to support the appropriate severity of illness, expected risk of mortality, and complexity of care," according to AHIMA.

In today's health care environment, the responsibilities of CDI and other documentation specialists are fluid. "AHDI expects further new roles to develop for people with health care documentation skills as technology changes and workflows evolve to accommodate those changes," Dooley says.

— ESR