Prospective, retrospective, and in-home programs are compared on RAF yield. Cost per accepted capture is the number that compares them honestly.
You know what you pay your chart review vendor. You know the rate on your in-home contract and what your internal coding team costs fully loaded. What almost nobody can produce on request is the risk adjustment cost per closed gap, broken out by the program that closed it.
It is not a hard calculation. It is an unowned one. The spend sits in three or four budgets, the yield sits in a risk adjustment report, and no one in the organization is accountable for dividing the second into the first. So programs get compared on RAF produced, which rewards whichever one touches the most members, and renewal decisions get made on a number that cannot distinguish an efficient program from an expensive one.
The comparison is worth forcing this year specifically, because one of the three programs is about to become structurally more expensive per unit of captured revenue.
The Denominator Is Wrong in Most Attempts
When organizations do attempt this, they divide program cost by gaps closed. That overstates the performance of every program and overstates some far more than others.
A closed gap is a condition your team confirmed and coded. It becomes revenue only when it lands on an encounter CMS accepts. Between those two states sit rejected encounters, diagnoses filtered as not risk-eligible, and voids that remove a diagnosis quietly enough that nothing in internal reporting registers a loss.
So the denominator is accepted captures, not closed gaps, and the distinction is not academic. Whether a condition reaches the risk score depends in part on how it was captured. A condition documented at a live visit generates its own encounter. A condition identified outside of one depends on an encounter that has to already exist, which is a dependency the closed-gap count does not show. Two programs with an identical cost per closed gap can differ on cost per accepted capture, and only the second number is describing revenue.
If your organization cannot yet produce accepted captures by program, that gap is itself the finding, and it means every program comparison made so far has been built on the wrong denominator.
What Belongs in the Numerator and Usually Is Not
The vendor invoice is the visible cost and frequently the smaller half.
Retrospective programs carry chart retrieval, internal coder time spent on second-level review, the analytics work to build and prioritize the population, and submission and correction labor on everything that fails. Much of that sits in operations rather than in the risk adjustment budget line.
In-home programs carry scheduling and member outreach, the no-show rate applied against a fixed per-visit price, clinical staffing or vendor margin, and the downstream work of turning an assessment into a coded, submitted diagnosis. The visit cost is a fraction of the cost per captured condition once the outreach funnel is included.
Prospective programs carry a cost that is easy to leave out because it does not arrive as an invoice. Surfacing conditions at the point of care takes visit time, so the efficiency of that workflow is itself a cost variable and belongs in the number.
Every program also carries an audit cost that nobody assigns to it. Captures that cannot be traced to a document and a coder create defense work later and repayment exposure after that. Because RADV findings are extrapolated across a contract, that exposure is not proportional to the number of weak captures a program produced. It is proportional to what those captures imply about everything else.
Want to see cost per accepted capture for your own programs?
The Same Condition Has Three Different Prices
Take one chronic condition on one member and price its capture three ways.
Documented during a scheduled visit the member was already attending, the incremental cost is the workflow that put the information in front of the clinician. The encounter was going to be generated regardless.
Captured during an in-home assessment, the cost includes the entire outreach funnel that produced the visit, including everyone who did not answer, plus the visit itself and the coding that follows.
Found in a retrospective review in the fourth quarter, the cost includes retrieval, review, coding, submission, and whatever correction work the submission requires, against a closing window that removes the option to try again.
Same condition, same member, same RAF value. The spread between those three prices is larger than most organizations assume, and it is the entire argument for moving detection earlier rather than adding capacity at the end.
Why the Retrospective Number Changes in 2027
Retrospective programs have carried a hidden subsidy: the chart review has been able to capture conditions independently, which means the cost of finding something late has been roughly the cost of finding it.
The CY 2027 Rate Announcement ends unlinked chart reviews beginning with the 2027 payment year, so a chart review diagnosis has to tie back to an encounter CMS accepted. The July HPMS memo on CY 2027 risk adjustment implementation sets out what has to be in place. For a condition with a qualifying encounter already present, the economics are roughly unchanged. For a condition without one, the path to payment is gone, which means the spend that found it produced nothing.
That does not make retrospective review worthless. It makes its cost per accepted capture dependent on a variable most organizations have never examined: how many retrospectively found conditions have a qualifying accepted encounter behind them. Anyone renewing a chart review contract for 2027 without looking at that is pricing the contract on last year’s economics.
V28 tightens it further. Fewer conditions carry payment, so the yield side of every program is smaller while the cost side is not.
How to Build the Number
Take one closed payment year and one population. For each program, assemble total cost including the categories above, then count the conditions that program produced which CMS actually recognized, confirmed against MAO-004 and the MOR rather than against internal capture reporting.
Divide. Then run it a second way, restricted to conditions that were both accepted and traceable to a document and a coder, which approximates the captures you would still hold after an audit. The distance between those two figures is the part of each program’s output you are not certain you keep.
Most organizations find the ranking changes when the denominator does, and that the program with the best cost per closed gap is not the program with the best cost per defensible accepted capture.
How Invent Health Makes It Measurable
Invent Health runs risk adjustment as one connected loop rather than three tools that integrate, and it starts with analytics, not coding, which is what makes cost per accepted capture calculable rather than estimated.
Risk Analytics decides what is worth working. Suspect logic runs across historical, lab, pharmacy, comorbidity, and chart-derived signals, drawing on EMR data through CCD and FHIR. Conditions are ranked by how much they move the score and how likely they are to close, which is the ranking that changes the cost side rather than the yield side, and internal capture is validated against CMS-recognized HCCs in the MOR with RAF progression tracked across payment cycles in the MMR.
The Coder Workbench confirms the condition is real. AI-assisted coding reads the clinical documentation, shows the supporting evidence, and recommends ICD-10-CM codes for a certified coder to confirm and sign. Invent Health’s NLP detects that supporting evidence with 85 percent or higher accuracy out of the box, and every code stays linked to the chart for audit-defensible lineage, which is what allows the second calculation above to be run at all.
Encounter Submissions makes sure it counts, validating 837P, 837I, and DME for Medicare EDPS and Edge Server XML for ACA, flagging missing and non-risk-eligible diagnoses before a file leaves, then reconciling MAO-002 and MAO-004 for Medicare and the Edge Server reports for ACA. A diagnosis that did not count returns as a known condition needing a corrected encounter rather than reappearing as a fresh suspect, which removes one of the larger hidden costs in the numerator.
Because the three run on one platform, each accepted capture can be attributed back to the program that produced it, which is the join most organizations are missing when they try to build this number by hand. For organizations working with delegated groups and IPAs, the same attribution rolls up to the group level.
Run It on a Closed Year
Pick a payment year that has finished reconciling and build cost per accepted capture for each program you fund. If the ranking differs from what your RAF yield reporting has been telling you, the renewal conversation you are about to have is the wrong one.
Frequently asked questions
What is cost per closed gap in risk adjustment?
It is total program cost divided by the conditions that program captured. The more useful version divides by conditions CMS actually accepted, since a closed gap only becomes revenue once it lands on an accepted encounter and loss rates between closure and acceptance differ by program.
How do you calculate risk adjustment cost per closed gap?
Take one closed payment year, total each program’s full cost including retrieval, outreach, coding, submission, and correction labor, then divide by the conditions that program produced which CMS recognized in the MOR. Using internal capture counts instead overstates every program.
Why should the denominator be accepted captures rather than closed gaps?
Because conditions are lost between capture and acceptance through rejected encounters, diagnoses filtered as not risk-eligible, and voids. Those losses are invisible in internal reporting, so cost per closed gap can rank programs in an order that revenue does not support.
What costs get left out of risk adjustment program comparisons?
Chart retrieval, member outreach against no-show rates, internal coder time, analytics and prioritization work, submission and correction labor. Most sit outside the risk adjustment budget line, which is why the vendor invoice is often the smaller half of the real cost.
Is prospective or retrospective risk adjustment cheaper per gap?
It depends on the population and the program, which is why the calculation is worth running rather than assuming. Prospective capture uses an encounter that is generated anyway, while retrospective capture adds retrieval, review, and correction work against a closing window.
How should in-home assessment programs be costed?
By including the full outreach funnel, not just the per-visit rate. Scheduling, member contact attempts, and no-shows all sit in the cost of each completed visit, and the coding and submission work that turns an assessment into an accepted diagnosis belongs in the number as well.
Does the 2027 chart review change affect program costs?
It changes retrospective economics specifically. Starting with the 2027 payment year, chart review diagnoses must tie back to an accepted encounter, so conditions found late without a qualifying encounter no longer have a path to payment. Spend that finds them produces nothing.
What should a plan consider before renewing a chart review contract?
How many retrospectively found conditions have a qualifying accepted encounter behind them. Without looking at that, the contract is being priced on economics that end with the 2027 payment year, and the yield assumption behind the renewal may not hold.
How does audit exposure factor into program cost?
Captures that cannot be traced to a document and a coder create defense work and repayment risk. Because RADV findings are extrapolated across a contract, that exposure is not proportional to the number of weak captures, which makes defensibility a cost variable rather than a compliance topic.
Why does the same condition cost different amounts to capture?
Because the cost depends on what had to happen to capture it. Documented at a visit the member was already attending, the incremental cost is workflow. Captured in-home, it carries the outreach funnel. Found in a fourth-quarter review, it carries retrieval, review, and correction against a closing window.
Does V28 change risk adjustment program economics?
Yes, on the yield side. V28 removed a large share of diagnosis codes from risk-adjustment mapping and re-based values, so fewer conditions carry payment while program costs stay flat. Every program’s cost per accepted capture rises unless its efficiency improves.
How does Invent Health support cost per accepted capture measurement?
Analytics, coding, and encounter submission run on one platform, so each accepted capture can be attributed to the program that produced it and validated against CMS-recognized HCCs in the MOR. That attribution is the join most organizations lack when building this figure manually.
