Every claim depends on two coding systems working together: CPT describes what was done, and ICD-10-CM describes why. Get the “why” wrong — too vague, missing a required detail, using a code the payer no longer accepts — and even a perfectly documented, medically necessary service can come back denied. (See the glossary for CPT, HCPCS, and other coding terms, and our guide to RCM for how coding fits into the bigger revenue cycle.)
This guide covers how ICD-10-CM codes are actually built, the coding-guideline concepts that trip people up most often, and — timely as of this writing — what’s changing in the FY2027 annual update taking effect October 1, 2026.
How an ICD-10-CM code is built
- Characters 1–3 (the category): A letter followed by two numbers, identifying the general disease or condition category. This is the broadest level of the code.
- Character 4 onward (etiology, anatomic site, severity): After the decimal point, additional digits add clinical detail — which side of the body, which specific site, what stage or severity. This is where “unspecified” and “specified” versions of the same condition diverge.
- Placeholder “X”: Some codes reserve a character position for future expansion. When a code requires a 7th character but doesn’t use every character in between, “X” fills the gap so the 7th character lands in the right position.
- 7th character extension: Common on injury and some obstetric codes, this final character indicates the type of encounter — “A” for an initial encounter (active treatment), “D” for a subsequent encounter (routine healing), and “S” for a sequela (a condition arising as a direct result of an earlier injury, coded after the acute injury has resolved).
Guideline concepts that cause the most denials
Specificity. ICD-10-CM almost always offers a more specific code alongside a vaguer “unspecified” one. Payers increasingly treat unspecified codes as a red flag, since they can suggest the diagnosis wasn’t actually pinned down — and an unspecified code is exactly the kind of thing that can trigger a CO-11 “diagnosis doesn’t support the procedure” denial if the documentation doesn’t clearly support the level of detail billed.
Laterality. Many codes require specifying left, right, or bilateral. Missing or mismatched laterality (a right-side diagnosis paired with a left-side procedure code) is one of the more common, avoidable rejection triggers.
Excludes1 vs. Excludes2 notes. An Excludes1 note means the two conditions cannot be coded together under any circumstances — they’re mutually exclusive by definition. An Excludes2 note means the excluded condition isn’t part of the code being reported, but the patient could have both conditions at the same time, in which case both codes can be used together. Confusing the two is a subtle but common coding error.
“Code first” and “use additional code” notes. Some codes require sequencing — a note might instruct that the underlying condition be coded first, with the current code reported second (or vice versa). Getting the sequence wrong can affect how a claim is adjudicated even when both codes are individually correct.
The annual ICD-10-CM update takes effect October 1 every year, and FY2027 brings 190 new codes, 30 deletions, and 4 revised code titles. The largest changes land in injury and poisoning (roughly 60 new codes, expanding toxic-effect coding), pregnancy and childbirth (about 44 new codes, adding site specificity for ectopic pregnancies), and musculoskeletal coding (around 31 new codes adding anatomical detail and laterality). Cardiac coding sees a real structural change too: dilated cardiomyopathy (I42.0) moves from a billable code to a non-billable parent code, meaning claims using the old code alone will need a more specific replacement. A handful of new Z-codes also add exposure-history and BMI-range options. The practical risk isn’t the new codes — it’s a deleted code still sitting in an active charge template or superbill past October 1, which generates denials from day one.
Common ICD-10-CM coding mistakes
Beyond the guideline concepts above, the errors that show up most often in practice: defaulting to an unspecified code out of habit rather than checking whether the documentation actually supports a more specific one; missing the 7th character extension on injury codes entirely, or using “A” (initial encounter) well past the point where treatment has become routine follow-up care; overlooking an Excludes1 note and submitting two codes that can never legally be reported together; and continuing to use a code that was deleted in the most recent annual update, simply because it was still saved in a template from the previous year.
Before October 1 each year: audit your practice’s saved charge templates, superbills, and EHR favorites lists against the new code set, and specifically check for any codes marked deleted in the update. This one habit prevents the single most common FY-update-related denial — billing a code that simply no longer exists.
Risk-adjustment coding leans heavily on ICD-10-CM specificity — see our ICD-10-CM Coding Guide for the coding fundamentals that carry over directly into HCC work, and our beginner’s guide to HCC coding and risk adjustment for how those diagnoses translate into Medicare Advantage payment.
Auditing your own coding accuracy is the natural next step after learning the ICD-10-CM fundamentals above — see our step-by-step guide to internal coding audits.