The problem
the gap after
The mistake is rarely the decision. It is what happens after.
Hospital software is very good at recording what was decided. It is much worse at showing whether it actually happened - whether the prescription was sorted out, the referral came back, the worrying test result was acted on, or the patient ever found a regular doctor.
So R47 starts somewhere else. ANCHOR treats each thing that was promised as an item in its own right: who owes it, what counts as done, when it is late, and a clock running the whole time. Once every promise is written down that way, it stops being a vague sense that something slipped and becomes something you can actually see.
What ANCHOR tracks
six promises
Six kinds of promise
Each one can be opened, given an owner, timed, missed, and put right - which is what makes it countable.
- 01
Medication mix-up
The list of drugs does not add up, and nobody has resolved it.
- 02
Worrying test result
A result came back abnormal and no one has acted on it.
- 03
Referral
A patient was sent to a specialist, and nothing has come back.
- 04
Contact after discharge
Someone was meant to check in after the hospital stay.
- 05
Something in the way
Cost, transport, language or housing is blocking the next step.
- 06
Teach-back
The patient was meant to be shown how to manage at home.
Why now
why this year
The weeks after discharge are now a payment problem, not only a care problem.
Hospitals have always discharged patients into a list of follow-ups. What changed is who pays when that list stays open.
Under CMS's Transforming Episode Accountability Model (TEAM), selected hospitals are accountable for cost and quality from surgery through the thirty days after discharge for five surgical episode types. The model runs from 1 January 2026 through 31 December 2030, and participation is mandatory for the hospitals CMS selected.
At the same time, FDA published final guidance on Predetermined Change Control Plans for AI-enabled device software (18 August 2025) - a clearer path for software that will keep improving after clearance.
- CMS TEAM: mandatory episode model; surgery through 30 days post-discharge; 1 Jan 2026 - 31 Dec 2030. Hospital headcounts are omitted because CMS participant lists move.
- FDA final PCCP guidance for AI-enabled device software functions, issued 18 August 2025. Guidance is nonbinding.
Why software
not only people
Hiring more navigators does not automatically close the loops.
Transition programs that put skilled people on the phone can help individual patients. Scaling that reliability across every discharge is a different problem.
A large 2025 randomized trial of a navigator-led telehealth sepsis recovery program found no reduction in the combined outcome of readmission or death at ninety days versus usual care, even though about two-thirds of patients in the program arm engaged at least once. That is not a verdict on every care team - it is evidence that good intentions and human effort do not, by themselves, guarantee closed loops at scale.
- ENCOMPASS / STAR (JAMA Internal Medicine, 2025): 90-day composite readmission or death 48.2% vs 48.0%, aOR 1.05 (0.90-1.24), p = .53; n = 3,548; 66% engaged at least once. Sepsis survivors in one health system - not all navigation, not all conditions.
ANCHOR exists to make each discharge promise visible, owned, timed, and chaseable - so follow-through is measured and orchestrated, not hoped for.
The evidence
already measured
The problem is measurable, and someone has already measured it.
R47 did not invent this problem, and is not the first to notice it. These are federal statistics and published research. They describe the gap ANCHOR is built for.
- 14.0
- readmissions per 100 adult admissions
- AHRQ, 2020. Adult stays. Includes the pandemic year.
- 17.0
- per 100 for Medicare patients
- AHRQ, 2020. Two million readmissions.
- $17,700
- average cost of one adult readmission
- AHRQ, 2020. Cost to the system, not price charged.
- 21%
- lower readmission risk with outpatient follow-up
- 2024 meta-analysis. Four conditions only, and heterogeneity was very high.
Roughly one adult hospital stay in seven is followed by another within a month, and each one costs about $17,700. For Medicare patients it is closer to one in six.
The last figure is the interesting one, and the one to be careful with. A 2024 review found patients who got an outpatient follow-up were readmitted less often - about a fifth less. But it looked at four conditions only, the studies disagreed with each other a great deal, and an association is not proof that the follow-up caused the improvement.
That is the honest shape of the opportunity. Something as ordinary as making sure the follow-up actually happens looks like it matters. Nobody has proven it end to end, and that is the gap ANCHOR exists to close.
These are 2020 figures because 2020 is the most recent year AHRQ has published an all-cause readmissions brief for. The database runs to 2022, and AHRQ has published from it for single conditions, but not for readmissions overall. Rechecked 4 September 2026.
Next
see it