Both directions of error are costly
Over-coding creates compliance exposure, because the provider remains responsible for the accuracy of claims submitted under their NPI. Under-coding leaves earned revenue uncollected for work that was actually performed and documented poorly. Audits that look only for over-coding miss half the problem, and practices that focus only on capture miss the risk.
Specificity in diagnosis coding
ICD-10-CM expects diagnoses to be coded to the level of specificity the documentation supports, including laterality and encounter type where the code set provides them. When a note records a condition without the detail the code set distinguishes, the coder must fall back to a less specific code, which can affect medical necessity support for the associated procedure.
Evaluation and management leveling
For office and outpatient E/M services, the AMA revised the guidelines effective January 2021 so that level selection is based on either the level of medical decision making or the total time spent on the date of the encounter. History and examination are still documented as clinically appropriate but no longer drive the level for these services. Notes that were structured for the older framework often under-support the level actually warranted, because the elements that now matter are not clearly recorded.
Time-based documentation
Where time is used as the basis for level selection, the total time on the date of the encounter needs to be documented in a way that makes the basis clear. A note that references time vaguely does not support time-based selection, and the service falls back to medical decision making whether or not that reflects the work performed.
Modifier support in the note
Modifiers convey that a service was altered by a specific circumstance without changing its definition, such as a distinct procedural service or a service separately identifiable from another performed the same day. The circumstance justifying the modifier has to be visible in the documentation. A modifier appended without narrative support is difficult to defend if the claim is reviewed.
What a compliant query looks like
A compliant query asks the provider to clarify ambiguous, incomplete, or conflicting documentation without suggesting a particular answer or leading toward a higher-paying code. It presents the clinical indicators found in the record and asks the provider to clarify, and it is documented as part of the record. Queries that offer a preferred answer are not compliant and can make the resulting claim harder to defend rather than easier.
What an audit should actually measure
A useful coding audit compares coded encounters against the underlying documentation and reports patterns rather than isolated errors. Recurring findings usually point to a documentation habit that can be corrected once, such as a template that omits a field the code set requires, or a workflow that records time inconsistently. Reporting the pattern back to the provider is what prevents the same finding recurring next quarter.