When a DRG downgrade lands in the underpayment workflow

The classification can send the team down the wrong path

I have seen sepsis-related DRG downgrades enter an underpayment workflow while the denials team had little visibility into the pattern. Follow-up focused on the difference between the billed DRG and the paid DRG. Reading the actual letters revealed that the payers were challenging the clinical support for sepsis.

That distinction matters. A contract calculation dispute, a coding dispute, and a clinical-validation dispute can all produce a payment difference. They require different evidence and different expertise. Repeating the billed DRG in an appeal does little to address a payer's stated concern about the diagnosis supporting it.

I also would not assume that the payer's conclusion is correct. The investigation has to establish what the record supports, what the payer actually challenged, and which response is appropriate. CDI, coding, physician advisors, and clinical appeal resources belong in that discussion.

What I would request first

I would start with a defined inpatient population and enough history to follow an account through its outcome. The initial request would include:

• Billed and paid DRGs, payer and plan, payment and adjustment activity, and relevant claim dates.
• Denial and variance records, reason and remark codes, workqueue assignments, and recorded root causes.
• The payer letters, appeal submissions, responses, and final account disposition for a targeted sample.
• The operational definitions used by the underpayment, denials, CDI, and reporting teams.

The letters are essential. Remittance codes and report categories may be too broad to explain a clinical dispute. I would reconcile the structured data with the correspondence rather than treat either source as complete on its own.

How I would investigate the Epic workflow

In Epic, I would review how the organization creates and classifies denial and variance follow-up records. I would compare Source Area, Owning Area, Root Cause, and Clinical Root Cause with the findings in the letters. Those fields answer different questions: where the issue originated, who takes the next step, and what actually caused the dispute.

I would ask the application team for the relevant workqueue configuration and rule records, including a CER extract where rules drive the selection or routing. I would want the rule context, criteria, property parameters, and references showing where each rule is used. Then I would test representative accounts to establish whether the configuration explains the routing we observed.

A shared reason code does not justify moving every related balance to a clinical team. I would define a bounded handoff for cases that need clinical review, identify who can make the determination, and confirm how the outcome returns to billing and follow-up. Changing an Owning Area label is only useful if the operational handoff and system behavior also work.

The prevention action belongs upstream

Once the pattern is understood, I would bring the findings to the people who own the documentation and coding workflow. A standardized SmartPhrase or SmartText can prompt the clinician to capture relevant evidence and clinical reasoning more consistently. SmartLists can structure choices, and appropriate SmartLinks can bring chart information into the template.

The template does not establish the diagnosis. Clinical owners need to approve the content, and a CDI query process needs to handle records that require clarification. I would also separate requiring a response inside a template from requiring that template to appear whenever a diagnosis is entered. Those are different build decisions with different workflow effects.

The objective is a record that accurately supports the clinician's assessment and a follow-up process that addresses the actual dispute. It is not to add identical text to every sepsis encounter or assume all payers use the same criteria.

How I would measure whether it worked

I would keep the case definition consistent before and after the change, including cases that previously lived in underpayments. Otherwise, reclassification can create a misleading trend.

The conclusion should explain whether the organization prevented new denials, recovered older balances, or reduced final losses. Counting the same amount as both an avoided denial and an avoided write-off overstates the benefit. A useful result connects the clinical finding, the workflow change, and a financial measure the team can defend.