Audit Defense Doesn’t Start When the Audit Starts
The strongest RADV defense isn’t built during the five-month response window. It’s built at the moment a coding decision is made. When a coder reviews a chart, validates the documentation against MEAT criteria, confirms the AI’s evidence mapping, and approves the code for submission, that decision, documented with its evidence trail, becomes the foundation of the plan’s audit defense for that specific HCC.
Plans that build evidence trails during coding don’t need to reconstruct them during audits. The defense already exists. When a sampled enrollee’s HCCs are pulled for RADV review, the team retrieves the evidence package that was created when the code was approved. The clinical note, the MEAT element mapping, the AI’s assessment, the coder’s validation, and the quality review are all documented and connected.
Plans that don’t build evidence during coding face a different reality during audit response. The code was submitted months or years ago. The coder who reviewed the chart may have left the organization. The clinical note exists somewhere, but the reasoning behind the coding decision was never documented. The team has to reverse-engineer the defense from scratch under deadline pressure.
The Evidence Chain That Auditors Follow
CMS auditors evaluate submitted codes by following an evidence chain. They start with the submitted diagnosis. They review the medical record submitted by the plan. They assess whether the documentation supports the diagnosis through MEAT criteria. They determine whether the condition was actively managed during the relevant encounter. Each link in this chain must hold for the code to be sustained.
When the plan’s coding process builds this chain at the time of the coding decision, every link is created under conditions that produce the best evidence: the chart is fresh, the coder has full context, the AI has mapped the documentation, and quality review has validated the assessment. When the chain is built during audit response, every link is created under worse conditions: the chart is stale, the coder may be unavailable, the reasoning is reconstructed rather than documented, and time pressure limits the quality review.
The difference in evidence quality between these two approaches directly impacts audit outcomes. Proactively built evidence chains produce stronger, more consistent defenses than reactively constructed ones.
Two-Way Coding as Preemptive Audit Defense
Programs that remove unsupported codes before submission are doing proactive audit defense. Every code deleted before it reaches CMS is a code that can’t appear in a RADV sample. Every unsupported diagnosis removed from the plan’s submissions reduces the plan’s aggregate error rate, which reduces the financial impact of extrapolation when it resumes.
Two-way retrospective review, where coders add supported diagnoses and remove unsupported ones, functions as continuous RADV preparation. Each review cycle cleans the plan’s submitted data. Each deletion strengthens the overall defensibility of the remaining code population. Over multiple cycles, the plan’s submission set converges on codes that are genuinely supportable, producing lower error rates when audited.
Where Defense Begins
The strongest position for any radv audit in risk adjustment is one where the defense was built before the audit was announced. Evidence trails created during coding. Unsupported codes removed before submission. Documentation quality validated at the point of decision.
Plans that invest in these upstream defenses spend the five-month RADV response window confirming and packaging their existing evidence rather than creating it from scratch. That’s the structural advantage that produces favorable audit outcomes.