Reviving an abandoned D-SNP risk adjustment program — and lifting RAF ~30% without adding audit risk

The situation

I stepped in to lead risk adjustment for a dual-eligible Special Needs Plan (D-SNP) of roughly 10,000 members. On paper, the plan had a risk adjustment program. In practice, it was one person mailing assessment forms and forwarding what came back. The plan’s average RAF sat at 1.07 — and for a duals population, that isn’t conservative coding, it’s a warning sign in the other direction. Duals are among the highest-acuity members in Medicare; a RAF near the program average meant the plan had almost no idea how sick its own members actually were.

What was actually wrong

The problem wasn’t aggressive coding or timid coding — the program simply wasn’t running, and no one could see clearly enough to know why. The providers who could document the population’s conditions had been handed a 13-page paper form for every member, with no reporting and no support, so they disengaged. The low RAF was the symptom; the disease was an operation that had stopped functioning, paired with a complete absence of the data needed to prove it.

What I did

I started where the leverage was: the providers and the data. We replaced the 13-page paper form with a simple submission of EMR output that could be automated, then I hired an analyst and built the performance reporting providers had never had — and reviewed it with them every month. Make participation easy, show people how they’re doing, and they engage.

That reporting also gave us eyes. For the first time we could measure the program, and the data surfaced a problem no one had known about: roughly 20% of members had bad or missing contact information and were never being reached. So I had a tool built that synthesized member, PCP, and pharmacy records into one working list. Callers validated each number and captured updates on the spot, and we reached 87% of those previously unreachable members — scheduling 65% for an assessment. Those assessments ran through an in-home assessment vendor whose performance I managed hands-on, inside their own call center, running weekly capacity calls against forecast and moving completion from roughly 20% to 30%.

The outcome

Over two years, average RAF rose from 1.07 to 1.39 — roughly a 30% increase in risk-adjusted revenue — not by coding harder, but by finally seeing and documenting a population that had been invisible. And it held: every RADV audit conducted during my tenure came back clean, because the increase reflected conditions that were real, present, and documented. None of this required exotic technology — the analytics ran on SQL, SAS, and Excel. It took operational discipline and a willingness to listen.

What this means for your plan

If your D-SNP or MA RAF looks low for your population’s acuity, the answer usually isn’t more coding pressure. It’s data you can actually see, providers you’ve made it easy to help, and vendors you hold accountable — with prospective coder review as the backbone that keeps every point defensible. Listen to your providers and make it easy for them to win. Treat every member as findable, because the ones you aren’t seeing are often the ones who need care most; in a duals population, that’s not just RAF, it’s the mission. And invest in deep operational and vendor oversight, measured weekly — that’s what separates a program that exists on paper from one that performs.