A risk adjustment supplier should be able to trace a suggested code to the supporting sentence in a patient record. It should also show who approved that code and explain the full cost of the service.
Evaluating these companies means checking clinical evidence, staff workload, data handling and contract terms together. Looking at any one of them in isolation can hide a problem in another.
In US Medicare Advantage, the Centers for Medicare & Medicaid Services (CMS) generally pays plans more for members expected to need costlier care.
That gives the documentation behind each diagnosis real financial weight. Australian Medicare works differently, so Australian readers should apply the buying method while assessing their own regulatory obligations.
Key takeaways
Decide whether you need help before appointments, after them or when records are selected for audit.
Run every supplier through the same cases and inspect the evidence behind each suggestion.
Include your own staff time and the cost of leaving when you compare quotes.
Define the job before building a shortlist
Start by deciding which of these three risk adjustment workflows needs attention.
Before and during appointments. Prospective support flags possible conditions for clinicians to assess and document during a visit. Any suggestion still needs clinical review.
After the appointment. Retrospective review examines completed records for diagnoses that were documented but never coded, or coded without supporting notes.
Preparing records for review. Audit support organises the evidence needed if CMS selects records for a Risk Adjustment Data Validation (RADV) audit. These audits check whether submitted diagnoses are supported by the medical record.
Before any demonstration, write down who will use the software and which problem you are solving. Note which electronic health record (EHR) system you run and whether you need software, people or both. The federal Health IT Playbook advises weighing a practice’s specific needs and resources when choosing an EHR, and the same principle applies here.
Suppliers package software and services differently. Navina builds its risk adjustment tool into the EHR for use at the point of care and says it integrates bidirectionally with patient records. Datavant offers first-pass coding along with second- and third-level review of previously coded charts.
Use supplier material to frame your questions. Then check every claim against your own requirements and what you see in the demonstration.
Make each vendor prove its claims
Prepare the same small set of synthetic cases (realistic but invented records) for every supplier. Purpose-built synthetic data generation tools can help create them without touching real patient records. Include one well-supported diagnosis, one suggestion that should be rejected and one case where two notes conflict.
Have a qualified clinician or coder set the reference answer before the demonstrations begin. Involve your privacy and legal advisers before any identifiable patient information leaves your systems.
Follow one recommendation. Ask the presenter to trace a suggested code to the original passage in the chart. Then ask to see the reviewer’s decision, the rejected suggestions and the exported evidence trail.
Reveleer says its Evidence Validation Engine pinpoints the relevant evidence in a chart and presents it to human coders for validation. Confirm any capability like this in a live demonstration rather than a slide.
Test the awkward cases. On 11 March 2026, the US Department of Justice announced that Aetna agreed to pay US$117.7 million. The settlement resolves allegations that Aetna submitted inaccurate Medicare Advantage diagnoses or failed to withdraw them. Aetna disputes the allegations and says the settlement is not an admission of liability.
The case shows why removal matters as much as capture. Ask the presenter to demonstrate how the software flags an unsupported code for deletion and tracks the correction.
For calendar year 2027, CMS has finalised the exclusion of diagnoses from chart review records that are not linked to an encounter. The exception covers members who switch from one Medicare Advantage organisation to another. It does not apply to people moving from Original Medicare into a Medicare Advantage plan, so ask each supplier how its workflow identifies these records.
Measure the work left behind. Record supported and unsupported suggestions, reviewer corrections and staff minutes per chart. Note any extra hand-offs the software creates.
When a vendor quotes an accuracy figure, ask how it was calculated and which cases were tested. Ask how much human review was needed to reach it.
For a structured starting point, RAAPID’s guide to top risk adjustment companies compares suppliers on audit defensibility, two-way coding and point-of-care fit. Use it to sharpen your demonstration questions, then score every supplier against the measures you agreed in advance.
Check who controls your patient data
Set minimum data-handling requirements before you compare features. Ask for a one-page diagram showing where records travel, where they are hosted and every subcontractor that handles them. It should also show who can access the data, how long it is kept and how it is deleted at exit.
Get written terms explaining whether the vendor may use your records to train its models.
Under HIPAA, the US health privacy law, covered entities need a business associate agreement (BAA) with any vendor that creates, receives, maintains or transmits protected health information on their behalf. A subcontractor doing the same work for that vendor is also a business associate.
A signed BAA sets obligations, but it does not prove the vendor meets them. The HHS cloud computing guidance states that its Office for Civil Rights does not endorse, certify or recommend specific products. That page was last reviewed in December 2022, so pair it with current legal advice when you assess a contract.
Private sector health service providers in Australia are covered by the Privacy Act regardless of annual turnover. A small practice is therefore not automatically exempt.
The OAIC received 1,205 data breach notifications in calendar 2025, an 8% increase on 2024 and the highest since the scheme began in 2018. Health service providers were the most affected sector, with 225 notifications. These figures span all organisations, so assess each supplier’s data practices on their own merits.
Compare the full cost of ownership
Compare quotes using the same modules, services and workload assumptions. Include every cost beyond the licence fee.
Ask for an itemised quote. Build a worksheet with four columns: cost item, included or extra, charging unit (per chart, per user or per month) and who is responsible.
List software modules, record retrieval, EHR integration, coding services, training, support, renewals and data export. Where modules are sold separately, confirm which prospective, retrospective and audit-support components the quote includes. Request current written prices and any minimum commitments from every shortlisted supplier.
Budget for your own team. Add internal hours for setup, training, review and changes to working practices. Confirm that support is available during your working hours.
Agree the exit terms. Establish the cost of recovering usable records and evidence trails at termination, in a format your next system can read. Ask for a reference from a deployment of similar size.
Conclusion
The right risk adjustment partner makes its evidence easy to inspect. Every suggestion should lead back to a sentence in the record and a named reviewer, and unsupported codes should be as easy to remove as new ones are to add.
Run every supplier through the same cases, agree your measures before the demonstrations and price the full cost of the relationship. Put acceptance criteria in writing before the pilot begins, so the decision rests on what each supplier showed you rather than what it promised.
FAQ
What is the difference between prospective and retrospective risk adjustment?
Prospective support flags possible conditions before or during a visit so clinicians can assess and document them. Retrospective review examines completed records for diagnoses that were missed in coding or coded without support.
What changes for chart reviews in 2027?
For 2027, CMS will exclude diagnoses from chart review records that are not linked to an encounter. There is an exception for members who switch between Medicare Advantage organisations, but not for people joining from Original Medicare.
Does a business associate agreement make a vendor HIPAA compliant?
No. A BAA sets out the vendor’s obligations, but it does not prove the vendor meets them. HHS also states that its Office for Civil Rights does not endorse or certify specific products.
Disclaimer: This article is general information only and does not constitute legal, compliance, financial or medical advice. The views expressed are those of the sponsor and do not necessarily reflect those of Dynamic Business.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.