What actually broke on January 4

• Jan 1 · NCCI PTP v33.0 effective. CY 2027 rates live. Telehealth modifiers BB and BC live. CMS-0057-F APIs due for impacted payers.
• Jan 4 · NCCI v33.0 implementation date and first real claim volume.
• CY 2027 performance period · The MIPS electronic prior authorization attestation begins.
• Coming · The next quarterly NCCI update lands in April. FY 2028 ICD-10-CM proposals surface at the spring CMS meetings.
• Dec 31, 2027 · Medicare telehealth flexibilities expire, absent Congressional action.

The week-two problem

Week one of January looked better than it should have. Volume was low, the encounters were simple, and the rejection count stayed inside normal range. [Insert your week-one rejection rate against your Q4 baseline.]

Week two was the real test. The complex charts people deferred over the holidays arrived together, and the rejections arrived with them. [Insert your week-two rejection rate.]

Any team that declared the transition clean on January 8 revised that view by January 15. If yours did, the lesson is about measurement timing rather than configuration: a change that lands on a holiday week does not show its true failure rate until ordinary volume resumes.

Separating the four causes

The diagnosis framework from November held up. For teams who wrote the failure-signature table in December, a January rejection sorted into one of four buckets within minutes:

• Rejections on code combinations, clustered by specialty · Cause: New NCCI edit pair · Fix: Coder briefing, usually self-correcting in a week
• Rejections on pairs billed cleanly for months, modifier attached · Cause: Modifier-indicator flip · Fix: Scrubber configuration, not training
• Payments posting at unexpected amounts, no rejection · Cause: Rate configuration · Fix: Billing ops, and check your posting rules
• Nothing at all · Cause: Telehealth BB/BC omitted · Fix: Nothing will tell you. Audit it on purpose.

Teams without that table spent [insert days] working out which change caused what. The thirty minutes in December was the highest-return thirty minutes of the quarter.

What the indicator flips actually did

[Insert your own findings. The structure below holds regardless of what you found.]

A modifier-indicator change from 1 to 0 fails in a specific way: the claim carries a modifier your coder has appended correctly for years, and the payer rejects it anyway. The rejection reason rarely says "indicator changed." It says the services are bundled, which reads to a coder like a new edit rather than a changed rule about an old one.

That misreading costs time. A coder who believes a pair is newly bundled stops billing it. A coder who understands the indicator flipped knows the service is still separately payable when the clinical circumstances justify it, and the question is whether this encounter qualifies.

If you found flips you missed in December, the useful follow-up is not retraining. It is asking how your team learns that a quarterly file changed a rule, because the answer is usually that nobody does until a claim fails.

Still silent: BB and BC

The telehealth modifiers produced no rejections, because they were never going to. That is the point and it remains the risk.

Run a deliberate audit this month. Pull your January telehealth volume, classify each encounter as direct, platform-contracted or incident-to, and check whether the modifier matches. The gap you find now is correctable. The gap a payer finds in 2029 is not.

What the configuration tax actually cost

Add it up honestly for your own operation. Hours spent pulling and filtering the NCCI change report. Hours cross-referencing it against your volume. Hours briefing coders. Hours loading and testing the scrubber. Hours in January diagnosing rejections rather than working them.

[Insert your number. Most teams land between two and four weeks of a senior coder, concentrated in the people they can least spare.]

That cost recurs every quarter, and the January quarter is only the largest instance. CODIO AI resolves NCCI edits, MUE limits and LCD/NCD coverage while the claim is being built, with quarterly updates applying automatically and nothing for your team to retrain on. A pair whose indicator flipped on January 1 resolves correctly on January 4 without anyone reading a change report.

Complete charts flow through AutoPilot for fully autonomous coding. Charts where the documentation is thin or contradictory stay in CoPilot, where certified coders review and finalize. Teams see 45% more output per coder, measured against a manual baseline: same team, same charts, before and after. Nothing runs autonomous on day one, and the lane widens only as accuracy is proven on your charts by your auditors.

Fill this section from your own January observations. Qualitative practitioner observation, labeled as such, never presented as measured data. Delete the section rather than publishing invented examples.

[Observation 1: what your coders noticed in the first ten days.]

[Observation 2: which specialty or payer produced the most week-two rejections.]

[Observation 3: anything that surprised your QA reviewers.]

April brings the next NCCI quarterly update. Same three categories, same dangerous one. The six-week plan from November works again, and the top-20-code-pairs list you built in December is reusable.

FY 2028 proposals surface in the spring. The CMS ICD-10 Coordination and Maintenance Committee meetings are where October 2027 gets decided. Reading proposals in spring beats meeting finished codes in September. Confirm the meeting dates on the CMS site.

The telehealth cliff is eleven months out. December 31, 2027. Planning that starts in Q4 is planning that starts late.

• [Article 1]
• [Article 2]
• [Article 3]