|
E-Newsletter • August 2026 |
Editor's E-Note
Artificial intelligence is rapidly becoming part of the clinical documentation process, helping providers improve efficiency and reduce administrative burdens. Yet as health care organizations embrace AI-powered documentation tools, concerns remain about hallucinations, context errors, and the impact inaccurate information can have on patient care, coding, compliance, and reimbursement. |
|
|
|
The challenge is that AI-generated documentation often appears polished and credible, making errors difficult to detect. As a result, health information management and clinical documentation integrity professionals are increasingly being called upon to ensure documentation remains accurate, compliant, and clinically defensible.
In this month's exclusive feature, we explore how AI hallucinations and context errors are affecting clinical documentation and why strong governance, auditing practices, and human oversight remain essential. As AI continues to evolve, responsibility for the integrity of the medical record still rests with people, not technology.
In addition to reading our e-newsletter, be sure to visit For The Record’s website at www.fortherecordmag.com. We welcome your feedback at edit@gvpub.com. Follow For The Record on Facebook and X, formerly known as Twitter, too.
— Dave Yeager, editor |
|
CDI’s Battle Against AI Hallucinations, Context Errors
By Elizabeth S. Goar
Following a checkup with his primary care physician, who used ambient listening tools to capture encounter notes, Medicomp System CMO Jay Anders, MD, was surprised to see diagnoses of atherosclerotic heart disease and diabetic complication listed in the clinical documentation. While he does have high blood pressure, he has never been diagnosed with heart disease, nor is there anything in his history that would account for the AI scribe inserting anything related to diabetes.
It’s a real-world example of how AI-generated documentation can be polished, structured, and compliant on its face, while silently misrepresenting what happened during the encounter. “That gap,” Anders says, “is where hallucinations live.”
Making the Distinction
Hallucinations occur when an AI system generates information that was never present in the source material. Clinical context errors can be more subtle. The information may exist somewhere in the record, but the AI interprets or applies it incorrectly.
“In my experience, clinical context errors can be particularly difficult to detect during human review because the individual facts may appear accurate on the surface. The problem is not necessarily that the AI invented information; it’s that it misunderstood the clinical context surrounding it,” says David Cohen, Greenway Health’s chief product and strategy officer.
|
Insurance Coverage for False Claims Act Exposure
By Grant E. Brown and Seán McCabe
According to the Department of Justice (DOJ), 2025 was a record-breaking year under the False Claims Act (FCA). FCA settlements and judgments exceeded $6.8 billion, more than doubling the $3.1 billion total for 2024 and far above the previous record of $6.2 billion collected in 2014. Whistleblowers filed 1,297 qui tam lawsuits, which also was a record, far exceeding the 980 suits filed in 2024. These amounts represent a significant upward trajectory from previous years, and we expect these trends to continue as Attorney General Todd Blanche has remarked that the DOJ “will continue to aggressively deploy” the FCA to achieve policy goals.
Research Reveals Opportunities for Millions in Previously Missed Revenue
Waystar, a provider of health care payment software, recently released "The State of the Mid-Revenue Cycle" report. The new research found unanimous interest among health care finance leaders in a single AI-powered platform connecting the mid-cycle to the final claim, yet 86% lack true integration between those systems, relying instead on manual data transfers or disconnected solutions across the revenue cycle.
|
Have a coding or documentation question? Get an expert answer by sending an email to edit@gvpub.com.
This month’s selection:
If a patient was admitted with acute liver failure but has other diagnoses that possibly could be caused by the liver failure, should they also be coded?
Linda Chase, RHIT
Director of Health Information
Gove County Medical Center
Response:
The short answer is yes, you potentially can. Just because the patient was admitted with acute liver failure doesn't mean you would not also code other comorbidities that were going on during the admission. If the provider documented them as a separate diagnosis, and they required treatment, evaluation, monitoring, etc, they may absolutely be reportable.
The one thing that I'd be really careful about is assuming the liver failure caused the other problems. Even if, clinically, we know something is likely caused by liver failure, we cannot assume that they're related unless the provider documented it, or if there's some sort of coding guideline that allows us to make the relationship.
Also, an area that's ridden with high error rates is hepatic and encephalopathy/coma because K72 codes already allowed for liver failure with or without coma. I highly recommend checking the index in tabular sections carefully before separating them into separate codes.
So, in summary, I would look at each additional problem, individually, and evaluate whether it is separately documented, is clinically significant during the stay, and is already considered part of the liver failure code.
— Shannon Cameron, MBA, MHIIM, CPC, is an accomplished leader in the health care operations and revenue cycle industry with 20+ years of experience in operational management with multistate, multidisciplinary health organizations. She possesses an in-depth, hands-on knowledge of end-to-end revenue cycles and a strong proficiency with federal regulatory policy and managed care, revenue integrity, and finance for both physicians and facilities across multiple specialties. |
Sandiola Introduces Sandpiper
Sandiola, a clinical documentation integrity (CDI) technology and services company serving community hospitals, has released Sandpiper, its purpose-built CDI AI prioritization engine that reviews 100% of inpatient encounters to ensure patient acuity is accurately captured, and the hospital is paid in full for the care it delivers. Sandpiper is trained on years of actual CDI cases and powered by Anthropic's most advanced models and a proprietary machine-learning algorithm. Sandpiper reviews every chart in under a second and delivers 100% case coverage 24/7/365. True to its "human at the helm" design, every Sandiola DRG recommendation is verified by the hospital's own CDI team and/or Sandiola's clinicians, each dual-certified in CDI and coding, so nothing reaches the medical staff that isn't clinically valid, compliant, and audit defensible.
Atropos Health and Health Universe Partner on Clinical Decision Support
Health Universe, the secure operating environment for agentic health care, has partnered with Atropos Health, the world’s largest creator of real-world evidence for clinical decision making. Health care organizations using Health Universe will gain seamless access to Atropos Health's clinical decision support capabilities directly within the platform, enabling clinicians to incorporate high-quality, patient-specific evidence into everyday care decisions. Together, Health Universe and Atropos Health give clinicians access to relevant guidelines, publications, and patient-specific insights within their existing workflows, while giving health systems the governance and transparency needed to deploy AI safely. |
| Advertising Opportunities |
Have a product or service you want to market to HIM professionals? Utilize the reach of For The Record to accomplish your marketing goals. Email our experienced account executives today at sales@gvpub.com or call 800-278-4400 for more information.
|
|
|
