E M Documentation

2026 CMS and Commercial Payer Crackdown on Emergency Department E/M (99281‑99285): MDM Level Justification, Time Documentation, and Downcoding Defense Strategies

When Your 99285 Becomes a 99283 Overnight

Last week, one of our client hospitals watched UnitedHealthcare knock five straight high-level ED E/M claims (99285) down to 99283. Same provider, same template, same ICD-10s. UHC’s “AI validation” flagged them for “insufficient risk element detail.” In plain terms: the algorithm didn’t see enough decision-making intensity to justify high-complexity MDM. That single downcode sliced each encounter from about $233 to $152 under the 2026 Medicare Physician Fee Schedule locality rate. Multiply that by 300 visits a month and you’re staring at $24,000 in lost revenue.

And no, this isn’t just one hospital’s headache. In 2026, CMS auditors and commercial payers like Anthem and Cigna have hardened their automatic E/M validation checks. It’s not only about documentation completeness anymore, it’s about pattern logic, what your data “looks like” compared to everyone else’s. According to RACmonitor, both regulators and payers are feeding machine learning models to spot “abnormal” MDM distributions. If you bill above the local average for 99285, you’re flagged before your coder even finalizes the charge. The bots know your tendencies better than you do.

MDM Hierarchy: How Payers Are Re‑Scoring Your ED Notes

The 99281‑99285 ladder has turned into a data science showcase. Algorithms analyze every MDM element, problems, data, and risk, against what they expect for that clinical story. If the note says “labs and imaging ordered” without connecting the results to the actual decision-making, systems like UHC’s default to "low data" complexity. CMS reviewers in 2026 pivot around diagnostic uncertainty and comorbidity interplay when deciding if something’s moderate (99284) or high (99285).

Right now, the same audit flags keep appearing:

  • Unclear clinical reasoning linkage: EKG ordered, but no line explaining what differential diagnosis it clarified.
  • Incomplete risk documentation: Patient treated for acute dehydration and sent home, but note fails to describe the severity or rationale for ED care.
  • Data quantity without interpretation: Listing five labs without explaining which results altered the plan.
  • Copied MDM from templates: Repeated language across patients, especially from auto-generated notes, triggers payer review scripts instantly.

Everyone knows 99285 equals high complexity, but your note has to sound like the provider actually wrestled with risk. “Reviewed labs and imaging” doesn’t cut it. Instead, give context: “Reviewed serial troponins to exclude NSTEMI in atypical chest pain.” That’s exactly what CMS’s automated logic parses for alignment with high-complexity MDM. The difference between paid and downcoded often comes down to one connective sentence.

Time Documentation: The Forgotten Tie-Breaker

Time has never been the star of ED E/M leveling, but this year it’s a strong safety net. Under the 2025 CMS revision, time supports but doesn’t replace MDM, it’s proof that the provider faced a complex case. A Cigna denial overturned earlier this year hinged entirely on time: the note showed 45 minutes of total provider time consistent with a high-acuity workup. Once the provider expanded on what that time covered, coordinating admission, counseling family, Cigna reversed the downcode.

Don’t stop at total minutes. Include how the minutes were spent: “30 minutes coordinating admission with cardiology.” That detail cuts through algorithmic assumptions and shows real complexity. And yes, those minutes have to be the provider’s own, no shortcutting by using nursing time.

Defending Against 2026 Downcoding Audits

Here’s the uncomfortable truth: first-pass downcoding now happens before any human review. As ICD10monitor’s coverage on machine-learning compliance shows, payers have trained their models to question documentation that “does not align with expected clinical complexity.” So traditional appeal strategies, those long arguments over individual charts, arrive too late. The fix has to start upstream, where the data gets generated.

Pull your E/M distribution reports. If your 99285 utilization tops 40% of unscheduled ED visits, expect scrutiny. Then retrain providers to document clear decision tiers. The three MDM parts, problem, data, risk, must all point in the same direction. If diagnostic breadth is wide but testing light, you’re in 99284. If multiple data points and high patient risk both show, you’re safe at 99285.

Also, watch your modifier use. When 99285 overlaps with critical care (99291), make the time break explicit. Anthem Blue Cross has been denying those combinations for “duplicative service” when the provider doesn’t clarify when MDM ended and critical care began.

Here’s what a strong defense note looks like:

“High-risk chest pain with differential of ACS, PE, and dissection. Labs, CTA chest, and serial EKGs reviewed personally. Patient required continuous cardiac monitoring pending CTA results. Risk: acute event with potential organ dysfunction.”

Brief, specific, and aligned with CMS 2025‑2026 E/M logic. No filler, just reasoning. The kind that algorithms can actually read without assumptions.

Monday Morning Fix: Audit, Tag, Retrain

Stop thinking denials are random events, they’re not. They’re machine judgments following mathematical patterns. So you have to beat them at their own pattern game. Do a random 20-note audit from your latest 99284‑99285 claims. Flag every missing link between data and reasoning. Teach providers to use subtle phrases AI keys on, “due to,” “to rule out,” “after reviewing results.” Sounds small, but it flips the machine logic.

Also, clean up your templates. Too many ED EHRs auto-fill “MDM: discussed with attending, patient stable for discharge.” That boilerplate screams low complexity. Delete it. Make space for honest narrative thinking.

The compliance world in 2026 has finally gone full algorithm, as RACmonitor keeps pointing out. And these systems don’t forget patterns, they amplify them. If your workflow doesn’t show structured reasoning in every MDM section, payers will keep classifying your toughest cases as “moderate” and shaving off that $80.

Monday’s action plan: Export your 99285 utilization by provider, line it up against payer denials, find the outliers, and build targeted MDM coaching for the top 10% variance group. Otherwise, UHC’s bots will deliver that coaching for you, one denial at a time.

Sources

Claims Assistant