The surge in high-acuity claims: What it means for health plans

As health plans continue to grapple with rising costs, declining profit margins, and regulatory pressures, coding intensity and ensuring appropriate reimbursement are top of mind. Coding intensity has increased across many care settings, and while it is often attributed to increasing patient acuity, one study suggests a more complex reality. Comparable conditions are increasingly being billed at higher levels over time—driven not only by clinical need, but also by industry shifts toward documentation templates that encourage more detailed coding, AI-powered clinical documentation tools that capture more comprehensive patient information, revenue cycle strategies designed to optimize reimbursement, and the increased familiarity with the nuances of E/M billing guidelines.

Regardless of intent, misalignment between clinical evidence, documentation, and billed complexity represents one of the most significant challenges for health plans, introducing significant compliance, operational, and financial risks. For many organizations, heightened regulatory scrutiny in 2026 is expected to further amplify that exposure.

Sepsis, in particular, remains a persistent pain point for health plans and providers. As one of the most clinically nuanced and frequently scrutinized conditions, sepsis is challenging to diagnose, document, code, and validate consistently—especially as clinical criteria continue to evolve. These complexities often lead to payer-provider disputes, retrospective audits, and downstream provider friction, even when care delivery itself was appropriate.

Fortunately, new AI-driven, clinically informed, transparent, and auditable solutions are enabling health plans to proactively validate coding intensity, address complex clinical areas like sepsis, reduce revenue leakage, and ensure compliance, while driving legitimate improvements in documentation quality.

The challenges and risks of coding intensity

Upcoding is commonly defined as billing for services at a higher level of complexity than the services actually provided or documented. However, the concept is often used more broadly to describe situations in which coding intensity increases beyond what would be expected in the absence of payment incentives. For example, the RAND corporation describes it in a slightly nuanced way: the coding of a patient to a higher complexity level than they would be if payment were unrelated to complexity. The Centers for Medicare and Medicaid Services (CMS) cited one example of upcoding as billing an established patient follow-up visit using a higher-level E/M code, such as a comprehensive new-patient office visit.

When coding intensity outpaces clinical evidence, health plan costs increase and compliance risk rises—particularly in clinically complex service lines such as evaluation and management, surgery, behavioral health, and sepsis.

Here’s a look at some of the most recent research:

• Between 2018 and 2023, the number of visits coded at higher complexity levels increased from 32.5% to 39.6% across emergency departments, urgent care, and physician offices, a 2025 study found.
• Between 2011 and 2019, upcoding practices in inpatient settings increased 41% and were associated with $14.6 billion in hospital payments in 2019, a 2024 study in the journal Health Affairs found.

While post-pay audits are conducted to catch issues, these are usually performed too late for errors to be reconciled, leading to provider disputes and unnecessary administrative burden. As a result, health plans are often left vulnerable to compliance risks, while providers face increased friction. Responding to audits requires additional time and resources, further straining operations on both sides.

In 2025, CMS announced that all Medicare Advantage plans would be subject to audits, leaving them vulnerable to financial and operational exposure and administrative burden. With increasing volumes of complex claims, ever-changing coding guidelines, and reliance on manual, paper-based processes, proactively validating coding intensity—particularly in areas such as sepsis—has become increasingly difficult.

The AI documentation clash

The rise of AI-powered provider clinical documentation tools has made the coding landscape even more complex. While these technologies can genuinely improve documentation quality and capture previously missed conditions, they can also inadvertently contribute to coding inflation through several mechanisms: 

• Improved documentation quality
  AI ambient scribes and documentation assistants help clinicians capture more comprehensive patient information in real time during patient encounters, enabling providers to focus more on their patients. This improved documentation becomes the clinical source of truth.
  
• Incidental phrase capture
  AI-powered note-taking tools may capture and document conversational phrases or mentions of conditions that aren't directly relevant to the current episode of care, potentially leading to higher-complexity codes that don't accurately reflect the primary reason for the visit.
  
• Clinical interpretation and evolving conditions challenges
  Providers using AI tools may document preliminary clinical impressions, such as suspected sepsis, that are later coded as confirmed diagnoses, even when subsequent clinical validation using objective measures, such as SOFA scores, does not support that diagnosis.

This presents a delicate challenge for health plans: how to ensure payment accuracy and prevent inappropriate coding without penalizing providers who are genuinely improving their documentation quality. The solution requires sophisticated, clinically grounded approaches that can distinguish between legitimate documentation improvements and coding inflation.

Identify and prevent coding intensity with Cohere Validate

With declining profit margins and rising regulatory pressures, health plans need a new approach to validate coding intensity and clinical support without creating unnecessary provider friction. The key is to implement solutions that adapt to the evolving documentation landscape while maintaining clinical accuracy and provider trust.

‍Cohere Validate is a proactive, clinically grounded, and transparent AI platform that reduces revenue leakage, improves efficiency, and maintains provider relationships by ensuring transparency and traceability to defined policies and criteria. With Validate’s Sepsis Audit Agent, the only fully-agentic AI solution that provides sepsis criteria configurability, health plans gain efficiency, improved provider experience, and optimized savings.

The solutions address upcoding challenges in the following ways:

Transparent clinical decisioning
‍Every AI-driven recommendation is backed by extensive clinical documentation, including cross-checking vital signs, lab results, and physician notes to ensure every billed diagnosis is fully supported. Our trained clinicians validate findings, exercise judgment, and constructively engage with providers. This combination of AI and human-in-the-loop expertise ensures accuracy and enables defensible audits, reduces friction in provider discussions, and fosters more trusted relationships.  

AI-driven accuracy 
‍The platform uses agentic AI, natural language processing (NLP), and machine learning (ML) to intelligently identify and validate claims using clinical evidence and cross-reference codes with supporting clinical documentation, with human-in-the-loop clinical and coding auditors. 

Code- and provider-level trend analyses
Monthly analyses identify unusual spending patterns or coding shifts. By monitoring upward trends in specific codes for individual providers—adjusted for the risk profile of their patient population—and benchmarking against peer provider cohorts, the platform can identify outliers and anomalies that warrant further review.

Clinical validation at scale
‍For complex conditions such as sepsis, the platform goes beyond documentation review to compute objective clinical scores, such as the SOFA score, and to validate whether diagnoses are clinically supported. This helps identify cases in which preliminary clinical impressions were coded as confirmed diagnoses without sufficient clinical evidence.

Expanded code coverage
‍While traditional payment integrity efforts focus on high-dollar claims—typically covering 7-9% of codes—Cohere's precision AI infrastructure enables cost-effective analysis across a broader range of codes, helping health plans address coding patterns that might otherwise go undetected.

Detection of “code-switching” behaviors
The solution monitors service category changes to flag when providers alter codes to bypass existing controls.

Cloud-scale analytics
Our cloud-based AI infrastructure allows for large-scale, detailed analysis that was previously cost-prohibitive, empowering health plans to detect anomalies earlier and at scale.

Augments human capabilities
AI-driven insights strengthen manual review efforts. By enabling consistent, accurate reviews, the solution allows health plans to maintain and improve provider relationships.

Pre-payment interventions
By surfacing anomalies before payments are made, financial leakage and administrative burden are reduced. Interventions enable health plans to shift from reactive loss recovery to proactive prevention and improved efficiency. 

‍Prevent and Address Upcoding in 2026
As health plans continue to face operational, financial, and compliance pressures, deploying new solutions that proactively identify and prevent upcoding through clinically grounded, transparent, and validated AI-driven insights will be critical.