The original version of this blog appears on Libman Education’s website here and was published on May 13, 2026.
Accuracy is one of the most talked-about metrics in coding and one of the least consistently defined.
Before you can govern accuracy, you need to agree on what it means. And that is harder than it sounds.
The American Health Information Management Association (AHIMA) and American Academy of Professional Coders (AAPC) both point to a 95% threshold as the standard, but 95% correct according to whose guidelines? The Centers for Medicare & Medicaid Services (CMS)? The American Hospital Association (AMA)? A specific payer’s reimbursement rules?
Most organizations default to CMS and AMA, but with payer influence growing on health system bottom lines, more organizations are starting to factor in payer-specific coding policies much earlier in the revenue cycle (right or wrong, it impacts revenue).
This matters because two organizations can both report 95% accuracy and be measuring completely different things. So, defining accuracy is the first priority. And in practice, the only reliable way to define it is through how you choose to measure accuracy. The methodology you use to calculate accuracy ultimately defines what “accurate” means in your organization.
There are three primary methodologies organizations use. Each is valid, but they are not interchangeable and they can produce very different results.
This approach looks at a batch of encounters and evaluates each individual code to determine whether it is correct. It is most commonly used in outpatient settings, where encounters tend to have fewer codes, but specificity matters more. This is especially relevant in areas like Hierarchical Condition Category (HCC) coding, where capturing the right level of detail directly impacts downstream risk adjustment and reimbursement. Because this method evaluates every individual code, it tends to produce the most stringent accuracy score. You are measuring at the most detailed level. In my experience, it is the most intensive approach, but it also gives you the clearest picture of true coding accuracy.
This method evaluates accuracy at the encounter level rather than the individual code level. A case is either correct or not based on whether the overall coding appropriately represents the encounter. This is most useful in payment models driven by groupers, such as Diagnosis-Related Groups (DRGs) or surgical Ambulatory Payment Classification (APCs). In these cases, the goal is to ensure that the full set of codes supports the correct reimbursement outcome. It is less granular than code-over-code, but more aligned to how certain types of encounters are actually paid.
Some organizations layer in weighted scoring, assigning more importance to certain codes over others. For example, primary diagnoses or Evaluation and Management (E/M) levels may carry more weight than modifiers or secondary diagnoses.
This approach can be useful in targeted or focused audits, especially when you are trying to evaluate performance in a specific area like (E/M) coding or a high-impact procedure. However, because it introduces weighting, it can skew your overall accuracy picture if used too broadly. You may end up with a strong overall score masking issues in certain code set areas. For that reason, it is often better used as a focused audit tool rather than your primary, day-to-day accuracy methodology.
These approaches are not interchangeable. More importantly, they can tell very different stories about performance. Your compliance team, your coding team, and your autonomous coding vendor all need to be measuring accuracy the same way. If not, you are not comparing apples to apples, and your accuracy results become difficult to compare. This is especially important when working with autonomous coding vendors. If their methodology does not match yours, you may think performance is improving or declining when, in reality, you are just looking at two different measurement models.
Your organization should clearly document:
If your organization has historically used case-over-case for inpatient and code-over-code for outpatient, carry that logic forward. Consistency matters more than chasing a new methodology. And just as importantly, make sure your vendor is aligned to that same approach.
Accuracy is not just a number; it is a definition. Define it early, document it clearly, and carry that methodology forward to your autonomous medical coding vendor. Because if accuracy is not defined correctly from the start, everything built on top of it becomes harder to trust.
Ready to learn more about autonomous coding? Check out Nym’s autonomous coding education hub for helpful articles, downloadable worksheets, and more!