BCBSA: AI coding tools contributing to $1 billion spike in healthcare costs

The perennial battle of the coding contingents is entering a new phase as AI-enabled medical coding tools become more popular with health systems.  

A new analysis by the Blue Cross Blue Shield Association (BCBSA) contends that AI-assisted coding is leading to a dramatic uptick in spending, as health systems start to bill hospital stays as more medically complex than before.

The report tracks a rise in patients being assigned into higher reimbursement categories due to severity of their conditions, resulting in approximately $942 million in additional spending for BCBS companies alone over the past two years.  

For example, between 2023 and 2025, claims at the highest level of complexity (MCC claims) for major bowel surgery increased from 20.2% to 22.7%, while non-complex cases declined from 36.6% to 32.8%.  

This shift added $60.8 million in incremental claims costs, as patients were moved into higher-severity, higher reimbursement categories – yet few received additional treatments that would be expected for a patient with increased risks.

Secondary diagnoses account for 70% of the added cost

Tactfully avoiding the highly charged phrase of “upcoding,” BCBSA says that its data shows a rise in “specific documented diagnoses driving severity escalation,” or “bump codes,” that are often derived from limited data.

An increase in secondary diagnoses accounted for 70%, or roughly $650 million, of the increased costs for Blue Cross and Blue Shield (BCBS) companies from 2023 to 2025, the Association says.

“Many of these diagnoses can be derived from single laboratory values or routine observations, making them particularly well-suited for detection by modern RCM technology,” the report says. “Technologies including ambient listening for observation codes and laboratory data mining appear to be contributing to this coding growth.”

This argues for “a clear disconnect between coding and treatment,” with more patients being tagged with higher complexity codes without any “evidence of corresponding change in care delivered,” BCBSA says.

“If patients are truly sicker, we’d expect to see more treatment,” said Luke Chalker, BCBSA’s senior vice president of product and data science. “For example, we’re seeing significantly more anemia diagnoses at these hospitals without a corresponding increase in transfusions. The disconnect between diagnoses and treatment suggests that AI is identifying more billable conditions, not sicker patients.”

Payers counter with AI tools of their own

But health plans aren’t taking the changes lying down.  While about 60% of health systems might be using AI for coding and revenue cycle management tasks, payers are also investing heavily in AI tools to assist with utilization review, including detecting aberrations in claims patterns that may indicate fraud.

It’s the next iteration of the constant back-and-forth between payers and providers, both of whom have incentives to hold their ground on coding.  As AI-assisted coding and claims review tools become more prevalent on either side, the tug-of-war (and a little bit of associated finger pointing) is likely to continue for the foreseeable future.  

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].