Nine days from now, on October 1, 2026, the FY 2027 ICD-10-CM code set becomes the only valid one for diagnosis coding on encounters from that date forward. CMS and the National Center for Health Statistics posted the files at the start of the summer: 190 new codes that can be reported, 30 codes that can no longer be reported, and four revised code titles. It is a smaller update than last year's 487 new codes, which is good news for training and bad news for attention, because a quiet year is when the cutover gets skipped.

The rule is the same every year and still trips practices. The date of service determines the code set. A visit on September 30 uses FY 2026 codes even if the claim goes out October 5. A visit on October 1 uses FY 2027 codes even if the note was started in September. Any code deleted on October 1 that appears on an October claim is rejected at the clearinghouse or denied by the payer, and a practice that updated its system on October 15 has two weeks of rework to do.

Key takeaways

  • 190 new codes, 30 deleted, 4 revised, effective for dates of service on or after October 1, 2026; the date of service, not the claim date, decides the code set.
  • Most practices will type a handful of the new codes: the adult BMI split (Z68.18, Z68.19), the cardiomyopathy codes (I42.01, I42.81), post-bariatric hypoglycemia (E89.830) and odontogenic sinusitis (J34.830 to J34.839).
  • The vendor loads the code table; the practice owns the superbills, favorites lists, problem lists, authorizations and quality measure logic.
  • The guideline changes are small this year, but the hypertension with heart disease and hypertensive crisis edits touch every primary care and cardiology coder.

Where the new codes are

By chapter, the additions cluster in three places. Chapter 19 (injury and poisoning) gains 60 codes, a large share of them toxic-effect codes for alkenes and cycloparaffins (T52.81 and T52.82) by intent and encounter; emergency, occupational medicine and toxicology will use them and almost nobody else will. Chapter 15 (pregnancy and childbirth) gains 44, driven by site specificity for ectopic pregnancy and 33 new codes under O31.4 for continuing pregnancy after vanishing twin syndrome, by trimester and fetus. Chapter 13 (musculoskeletal) gains 31, including laterality for plantar fascial fibromatosis (M72.20 to M72.22).

The codes independent practices will actually type are scattered and specific:

New codeDescriptionWho uses it
Z68.18 and Z68.19Adult BMI 18.4 or less, and adult BMI 18.5 to 19.9, splitting the old single code for adult BMI under 20Primary care, geriatrics, eating disorder programs, anyone reporting BMI for quality measures
I42.01 and I42.81Familial-genetic dilated cardiomyopathy; arrhythmogenic cardiomyopathy (with I42.00, I42.09 and I42.89 filling out the family)Cardiology, and primary care managing these patients
E89.830 and E89.838Post-bariatric hypoglycemia; other postprocedural hypoglycemiaEndocrinology, bariatric surgery, primary care
C78.31, C78.32, C79.83Secondary malignant neoplasm of larynx, of pharynx, of oral cavityOncology, ENT
J34.830 to J34.839Odontogenic sinusitis by sinus (maxillary, ethmoid, frontal, sphenoid, unspecified)ENT, oral surgery, primary care
M72.20 to M72.22Plantar fascial fibromatosis, by lateralityPodiatry, orthopedics, primary care
Z86.17Personal history of Clostridioides difficile infectionPrimary care, gastroenterology, infection control reporting

Two of these matter beyond the specialties named. The BMI split affects every practice that reports BMI-based quality measures, because the old code will be invalid and quality logic that references it needs updating. The cardiomyopathy codes matter for risk adjustment in Medicare Advantage and ACO populations, where specificity in the chronic condition affects the HCC that is captured and the documentation the payer will expect to see behind it.

The guideline changes worth ten minutes

The FY 2027 Official Guidelines for Coding and Reporting, published separately from the code files, carry one new guideline and edits to four existing ones, which is as light as it gets. Two of the edits are in the circulatory chapter and belong in every primary care and cardiology coder's briefing. Guideline I.C.9.a.1, hypertension with heart disease, now says a code from category I11 is assigned when the patient has one or more of the listed heart conditions, closing an argument coders have been having for years about patients with two of them. Guideline I.C.9.a.10, hypertensive crisis, adds I1A (resistant hypertension) to the "code also" instruction, so a hypertensive urgency in a patient with documented resistant hypertension now carries both codes.

Neither change creates a new code. Both change what a compliant claim looks like on October 1, and neither will be caught by the claim scrubber, because the scrubber checks validity, not guideline compliance. This is the part of the update that only training fixes.

A worked example: the BMI split

A family practice reports a BMI screening and follow-up quality measure to two payers and to MIPS. Its EHR template maps a calculated BMI under 20 to the old adult code. On October 1 that code is invalid. Three things happen if nobody acts. Claims for October visits carry an invalid diagnosis and reject at the clearinghouse, which the practice will notice within two days. The quality measure numerator logic, which counts a BMI code plus a follow-up plan, silently stops counting those patients, which the practice will not notice until the fourth-quarter report. And the problem list for underweight patients keeps the old code, which flows onto every future claim until each chart is touched.

The fix takes an hour. Update the template mapping so a BMI of 18.4 or less maps to Z68.18 and 18.5 to 19.9 maps to Z68.19, update the measure logic (or confirm the registry vendor has), and run the EHR's inactive-problem-code report to find the patients whose problem list needs the new code at their next visit. The hour is only cheap if it happens before October 1.

The nine-day cutover checklist

  1. Load the code set. Confirm with your practice management and EHR vendors that the FY 2027 tables are loaded and effective-dated October 1. Most vendors do this automatically; some require you to accept an update. Ask for the confirmation in writing.
  2. Run the deleted-code report. Search every superbill, favorite list, order set, problem list template and charge master for the 30 deleted codes. Each hit needs a replacement code from the conversion table CMS published with the update.
  3. Update problem lists that carry deleted codes. A chronic condition on a patient's problem list coded with a soon-to-be-invalid code will flow onto every October claim. Most EHRs can produce a list of patients whose active problems map to inactive codes.
  4. Check the claim scrubber. Send a test claim with an October 1 date of service and a new code, and one with a September 30 date and an old code. Both should pass. If the scrubber flags the new code as invalid, its table is behind.
  5. Check prior authorizations that straddle the date. An authorization issued in September with a diagnosis code that is deleted on October 1 may not match the October claim. Call the payer for the ones with high-dollar services scheduled in October.
  6. Update quality measure and registry logic. Anything that references a deleted code in a numerator or denominator definition. The BMI codes are the obvious example.
  7. Brief the coders and providers. A one-page summary for each specialty in the practice, with the new codes they will use, the deleted ones they must not, and the two hypertension guideline edits. Providers who select their own diagnoses from a favorites list need the list fixed, not a memo.
  8. Hold the rejection report on October 2. Someone reads the clearinghouse rejection report the morning after the first October claims go out, looking for invalid diagnosis code rejections. Anything found is a template you missed.

What we see go wrong

The vendor update that was scheduled and failed silently, discovered when 300 claims reject. The problem list that was never cleaned, so a deleted code appears on every visit for a patient until someone notices a pattern in denials. Coders who learned the changes but providers who select codes from a favorites list that nobody touched. And the reverse error: practices that load the new codes early and use them on late-September visits, which is a rejection or, worse, a paid claim with an invalid code that gets recouped later.

One more habit we'd encourage. Keep the FY 2027 conversion table in the coding folder all year. In January, when the CPT changes land, the same table tells you which diagnosis codes on your maternity and procedural claims also changed in October.

Questions we hear

Our EHR vendor says the update is automatic. Do we still need the checklist?

Yes. The vendor updates the code table. It does not update your superbills, your favorites lists, your problem lists or your authorizations. Those are yours.

What about the coding guidelines?

The FY 2027 Official Guidelines for Coding and Reporting are published separately from the code files. Read the sections for the chapters your practice codes most; the hypertension with heart disease and hypertensive crisis edits are the ones we'd start with in primary care and cardiology.

Where does training fit?

Our live courses cover the FY 2027 update in the October sessions, and practices using Revelrex coding receive the specialty summaries as part of the service. Practices doing it themselves should plan a one-hour session per specialty before Friday, October 2.

What to do this week

  1. Get written confirmation from the practice management and EHR vendors that the FY 2027 tables are loaded and effective-dated October 1.
  2. Run the deleted-code search across superbills, favorites, order sets and the charge master, and fix every hit.
  3. Pull the inactive-problem-code patient list and flag those charts for the next visit.
  4. Send the two test claims through the scrubber and read the result.
  5. Schedule the specialty briefings and the October 2 rejection report review, each with a named person.