Medical Coding, Meet Artificial Intelligence
The robot in the next cubicle may become your favorite coworker.
If you’ve trained Siri to call you Princess or talk to your Alexa like she’s one of your kids, you’re becoming a friend of artificial intelligence (AI). AI has been around for a few years, even affecting medical coding and billing, but it’ll have a greater impact on your career in the near future.
Using AI to Enhance, Not Replace
Provider groups, facilities, and payers are turning to AI to help handle all sorts of things — patient care, documentation, planning, and more. AI is the ability of computers to learn to act on the information they have, to evaluate and deduce. Real AI is a far cry from its popular image. Think more of a black box, like IBM’s Watson, rather than a murderous android from “Westworld.”
Watson and other supercomputers are helping providers and payers identify the most efficacious and efficient treatment options. Several medical centers, including Cleveland Clinic and MD Anderson, are not only feeding it information but participating in research to exploit AI to best help providers and their facilities better care for patients.
AI can modify a provider’s behavior. For example, in an experiment in Chicago, an electronic health record (EHR) system was taught to monitor patients’ conditions and shift appropriate patients’ care and medication schedules so they could sleep longer. Nurses and providers were trained to follow the EHR’s recommendation rather than regularly, but needlessly, wake patients. Patients who the EHR let sleep longer did better than those treated in the standard way.
A provider’s diagnosis of a new patient doesn’t only support medical necessity, it spins the wheels of payment, utilization management, risk adjustment, quality management and reporting, retrospective review, and discharge management. An error releases a wave of miscalculations that affects the facility, provider, patient, and payer. AI can catch the error, suggest a different code, or alert an auditor in real time.
Does that mean coders are in danger of being pushed out of their cubicles by black boxes? Probably not.
The EHR Craze and the Luddite Fallacy
Back in the late 1800s, skilled weavers watched English fabric manufacturers install steam-powered machinery to speed production and lower costs. The angry workers put their bodkins and shears aside and started roaming the English countryside, following Ned Ludd, and destroying the newly automated weaving racks in the factories. Despite their righteous rage at “deceitful practices,” some of the Luddite leaders were banished to Australia or hung. Followers adapted to working in their newly industrialized world.
The Luddites were on the wrong side of the Industrial Revolution, it appears, and they unknowingly made a fateful mistake. New technology usually doesn’t take away jobs or livelihood, but it can change them.
Coders faced this when the U.S. Department of Health and Humans Services (HHS) incented providers and facilities to install EHR systems. “Meaningful Use,” as outlined in the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009, was intended to force facilities and providers to adopt EHRs to make documentation, billing, and payment more standardized and portable. Federal financial incentives prompted a rush of EHR start-ups and companies to pitch, among other things, that medical coders and billers would no longer be needed with the new technology.
Hearing this during the Great Recession of 2008, many medical coders panicked, naturally. But the threat proved hollow. EHRs didn’t provide what was promised, and healthcare facilities and providers didn’t replace their staff with machines. More coders were needed to help set up the systems, audit claims, and help providers improve their clinical documentation. Many EHR providers folded or merged with competitors, and some were sued into nonexistence by angry providers when the systems proved ineffective or too unwieldy. New federal and commercial efforts to predict patients’ risk, based on the EHR data, spawned the risk adjustment movement, making even more jobs available. More coders were needed to audit the set-up and EHRs, monitor and teach physicians to document in the new technology and newly-implemented ICD-10-CM codes, and wrangle EHR data.
A decade later, the medical coding field has tripled in need. AAPC, for example, has gone from a membership of 73,000 to more than 180,000. The U.S. Department of Labor’s Bureau of Labor Statistics predicts our field will grow another 13 percent by 2026.
When technology threatens, it’s easy to panic, but we can’t commit what economists call “The Luddite Fallacy.” Workers assume there is only so much work, the way their work is done can’t change, and the future will be bleak, but this isn’t usually the case. Technology usually creates more work for those willing to adapt.
The answer lies in our ability to transition to changing technology. Here are some of the AI-based initiatives affecting the revenue cycle:
Computer-assisted Coding (CAC)
CAC is one of the oldest forms of AI in our field. The original intent was for the computer to automatically assign the codes and submit the charges. But like flying cars and teleportation, the idea hasn’t fully materialized; however, CAC does help assign modifiers, catch correct coding edits, identify errors, and other tasks, freeing medical coders and billers to concentrate on other things. CAC proponents now concede that the human element is vital to accurate medical coding and billing. This technology also helps speed the reimbursement cycle. CAC can be a lifesaver, just like that timer on your coffee pot.
Computer-assisted Physician Documentation (CAPD)
CAPD is being added to EHRs to help providers address gaps in their clinical information. The AI reviews documentation and then guides the provider to adjust the documentation to assure it properly reflects the patient’s condition. The technology also helps capture complications and comorbidities that may affect the patient’s care and payers’ risk now or down the road.
AI is the perfect tool to sort through old records to identify conflicting diagnoses, out-of-date medications, and other inconsistencies. This can also help identify quality measurement issues and protect the provider in the future.
Care Transition Analysis
This technology allows payers to identify suspicious or repetitively incorrect billing patterns, and they can use it to coach or exclude the provider or facility from participation. Large practices, facilities, and health systems will use this to self-identify problems in medical coding and billing. The data can help auditors and clinical documentation improvement staff create reviews and coaching to improve reimbursement cycles.
EHRs not only did not replace coders, they helped fuel demands for better clinical documentation. Providers soon found themselves entering information as they assessed their patients, and visits turned into “watch-the-doc-on-the-computer.” Scribes solve that by shadowing the provider during an office visit and inputting the needed information. They help assure the documentation is complete and ready for the EHR, and let doctors concentrate on their patients.
One Holy Grail of electronic data, however, is interoperability of all this technology. EHRs still don’t talk to each other very well, and the Centers for Medicare & Medicaid Services continues to encourage communication. As EHR manufacturers wrestle with standards, others are looking at new ways of doing it. Apple, for example, is experimenting with a way for patients to carry all personal health information with them from provider to provider, and the history, lab results, and other information can be pulled from an iPhone. Imagine the opportunities arising there!
Trust Your Inner Algorithm
None of this change comes without some personal disquiet and effort. Medical transcriptionists displaced by natural language programming (NLP) and EHRs are transitioning to medical coders, scribes, and related roles. Imagination, curiosity, and flexibility are key, AAPC members say.
Elcilene Moseley, CPC, CCS, admits she felt threatened when CAC was introduced to her work. “The idea of losing my job to a computer software wasn’t a pleasant one. But after using it for a few years, I’ve come to realize that coding is complex and multi-faceted and certain procedures are so complicated that it will take a long, long time before any AI can properly fully code a chart,” she said. Pointing to ICD-10’s complexity, she added, “I’m not sure if it ever will.
CAC has been a big help to Moseley. “One of its purposes was to help coders become more productive and accurate,” she said. “It has succeeded some. The software sometimes picks up diagnoses I might have missed and/or provides procedure codes that I may not be familiar with, making me research and learn about them.” She admits her CAC also picks up stuff it’s not supposed to. “It hasn’t quite mastered combination codes and bundling issues, guidelines, and the constant new changes we coders have to be on top of.”
One of the many ways to welcome your new AI coworker is to make good use of your lighter workload by partaking in activities, education, and credentials. Is your group, facility, or healthcare system expanding? Are there new roles you can take on or move to? Opportunities for improvement? How can you become a leader thanks to, or in concert with, your AI buddy?
“I’ve realized that my role as a coder is slowly, but surely, switching to an auditor one. I’m no longer threatened by the software. I’m enjoying auditing the codes put out by the AI, and I have also learned quite a bit from it. Future-wise, I see myself auditing more and more, as the software becomes smarter and learns from the code corrections I make. At some point, we coders will probably end up becoming auditors. We’re nowhere near it yet,” she said.
There are a lot of opportunities out there: new roles, credentials, and jobs. Don’t be a Luddite; take some time to identify and pursue those opportunities.
Moseley agrees. “I started out coding by paper, flipping pages, 12 years ago. Now it’s all mostly automated. It’s all about going along with the new changes, and to never stop learning.”
IBM, Watson: www.ibm.com/watson/health/index-1.html
The Luddites: www.historic-uk.com/HistoryUK/HistoryofBritain/The-Luddites/
The Luddite Fallacy: www.economicshelp.org/blog/6717/economics/the-luddite-fallacy/
Becker’s Health IT & CIO Report, How AI Can Improve Accurate Coding — 2 Experts Weigh In: www.beckershospitalreview.com/healthcare-information-technology/how-ai-can-improve-accurate-coding-2-experts-weigh-in.html
HHS, HITECH Act Enforcement Interim Final Rule: www.hhs.gov/hipaa/for-professionals/special-topics/hitech-act-enforcement-interim-final-rule/index.html
HealthWorks Collective, Augmenting Coding and Billing With AI: www.healthworkscollective.com/augmenting-coding-and-billing-with-ai/
The University of Chicago Medical Center, SIESTA Project Reduces Inpatient Sleep Interruptions: www.uchicagomedicine.org/forefront/patient-care-articles/2019/january/siesta-project-reduces-inpatient-sleep-interruptions