Tuesday afternoon, 3:47 PM. I’m reviewing audit results when my stomach drops. Forty-three percent of our submitted codes just failed validation. Not because we coded wrong conditions. Because our risk adjustment HCC coding was based on assumptions, not documentation.
We’d been coding “diabetes with complications” every time we saw metformin and an ACE inhibitor together. Made perfect sense clinically. Except nobody ever documented the actual complications. We assumed. CMS doesn’t pay for assumptions.
The Assumption Epidemic
Here’s what nobody admits: most HCC coding runs on educated guesses. We see chronic kidney disease medications, we code CKD. We see depression drugs, we code major depression. We see heart failure symptoms, we code CHF. Except that’s not coding. That’s medical detective work, and it fails audits every time.
Last month, I audited 500 of our highest-value HCCs. Two hundred and twelve were built on interpretation rather than documentation. That’s $2.4 million in risk adjustment revenue hanging on our ability to read between the lines. When CMS came calling, those lines disappeared.
The scariest part? Our accuracy rate was 96%. We were coding the right conditions. The patients absolutely had these diseases. But being clinically correct doesn’t matter if you can’t prove it from the documentation. We’d built an entire program on being right instead of being compliant.
Take Mrs. Patterson. Three nephrologists treating her. Clear kidney disease. Dialysis three times weekly. We coded CKD Stage 5. Seemed obvious, right? The audit failed because nobody ever wrote “chronic kidney disease stage 5” anywhere. They wrote “ESRD” and “renal failure” and “kidney issues.” Different words. Lost HCC.
The Connecting Dots Disaster
Coders aren’t supposed to be doctors, but we’ve trained them to act like it. They’re connecting clinical dots to find HCCs. Smart? Yes. Compliant? No.
Jimmy, one of our best coders, showed me his process. Patient has insulin, so probably diabetic. Also has neuropathy medications, so diabetes with neurological complications. Also seeing nephrology, so diabetes with renal manifestations. From a few medications and appointments, he’d built three HCCs. All probably accurate. None explicitly documented.
The provider documentation supported Jimmy’s conclusions if you understood medicine. But CMS auditors aren’t paid to understand medicine. They’re paid to verify exact documentation. When they couldn’t find “diabetes with neurological complications” written clearly, the code failed. Jimmy wasn’t wrong. The process was.
We’d turned coders into diagnosticians. They were so good at interpreting clinical signals that they’d stopped requiring actual documentation. The better they got at understanding medicine, the worse they got at compliant coding.
The Literal Documentation Revolution
We fixed our HCC coding with one brutal rule: if it’s not written exactly, it doesn’t get coded. No interpretation. No assumption. No connecting dots.
The pushback was immediate. “But the patient obviously has CHF!” Sure, but obvious doesn’t pay. “The medications prove diabetes complications!” Medications prove nothing without documentation saying so.
Our capture rate initially plummeted 30%. Coders were finding fewer HCCs because they couldn’t assume anymore. Revenue projections looked terrible. Leadership panicked. Then something interesting happened: providers started documenting better.
When we stopped accepting “cardiac issues” as CHF, providers started writing “chronic systolic heart failure.” When we rejected coding CKD from lab values alone, nephrologists began documenting stages explicitly. The pressure to document completely shifted from coders to providers, where it belongs.
The Copy-Paste Compliance Method
Here’s our new process, stupidly simple: coders must copy and paste the exact text supporting each HCC. Can’t find text to paste? Can’t code it.
Mrs. Patterson’s chart now shows: “Chronic kidney disease stage 5” copied directly from nephrology notes. Not interpreted. Not assumed. Copied. When auditors review, they see exactly what we saw. No detective work required.
This slowed coding initially. Coders complained about the extra step. Then audit results came back: 94% validation rate, up from 57%. The time spent copying text saved months of audit response and millions in penalties.
We built a library of acceptable documentation phrases for each HCC. Not for providers to copy, but for coders to recognize. “Type 2 diabetes with diabetic neuropathy” works. “Diabetes with nerve pain” doesn’t. Coders stopped translating and started matching.
Your Thursday Morning Test
Pull up ten of your highest-value HCCs from last week. For each one, try to copy and paste the exact documentation supporting it. Not paraphrase. Not summarize. Copy and paste.
If you can’t find explicit text for more than two, you’re coding assumptions. Your smart interpretations are future audit failures. Your clinical accuracy is compliant disaster.
Count how many codes require connecting multiple pieces of information. Every connection is a vulnerability. “Patient has A, and also has B, therefore C” fails even when medically obvious. You need documentation saying “C” directly.
The truth hurts: good HCC coding isn’t about understanding medicine. It’s about finding exact words. Your clinically sophisticated coding team might actually be your biggest audit risk. The coder who knows less medicine but requires explicit documentation will beat the medical expert who connects dots every time.
Stop rewarding interpretation. Start demanding documentation. Your audit results will thank you, even if your coders won’t.
For more, visit Pure Magazine