Reduce 30-day readmissions by contacting every patient, not the top decile.
OneDose reduces avoidable 30-day readmissions by removing the capacity constraint on post-discharge follow-up. Structured check-ins run on every discharge — Day 1, Day 7, Day 14, Day 30 — catching a prescription never collected, a side-effect that ended a therapy, or early deterioration, and escalating to a clinician while the problem is still small.
The intervention is not new or contested. What is new is being able to run it on everyone rather than on the patients a rota had room for.
Which readmissions can actually be prevented?
Not the ones where the disease progressed. Those patients were coming back, and no amount of contact changes that — a vendor implying otherwise is describing a product that does not exist.
The addressable group shares a shape: something small went wrong in the first fortnight, nobody noticed, and it compounded until it needed a hospital. The originating failure is usually mundane and usually about medication.
The medication was never collected
The prescription was written and never picked up — cost, transport, confusion about where to go, or a bad week. Nothing in a normal discharge process checks, so nobody knows.
It was collected and taken wrong
Wrong dose, wrong time, or stopped once the symptoms improved. Discharge instructions are given to someone who has just been in hospital and is not at their most receptive.
A side-effect ended it
The patient felt worse on the medication than off it, made a rational decision, and told nobody — because the person to tell was not obvious and the alternative was ringing a hospital.
Deterioration went unreported
A symptom appeared that the patient did not recognise as important. The threshold for "should I bother someone" is set by the patient, and it is usually set too high until it is an emergency.
Every one of those four is detectable by asking a patient a structured question at the right moment. Every one of them is invisible if nobody asks.
Why does the follow-up not happen today?
Not because anyone disagrees that it works. Because of arithmetic. A few hundred discharges a week, several contacts each, most unanswered on the first attempt, each answered call taking ten to fifteen minutes to conduct and document — that is a very large number of hours from nurses and pharmacists you cannot hire more of.
So every organisation does the same rational thing: stratify by risk, call the top of the list. Given a fixed number of hours that is the correct allocation, and nobody should be embarrassed by it.
It is nonetheless expensive at the population level, for a reason that is easy to miss. The highest-risk cohort is small and is already the cohort everyone is watching. The unmanaged risk sits in the medium-risk group — individually less likely to deteriorate, collectively far larger, and structurally unobserved because nothing in the process ever asks it a question.
What does OneDose actually do about it?
It changes the marginal cost of one more contact so that it is no longer a nurse-hour. Every discharged patient gets a follow-up plan derived from their discharge summary, prescription and risk profile. The agent makes the contact, structures the answers, and routes to a clinician only when an answer crosses a threshold, confidence is low, or the patient asks for a person.
The care team’s day changes shape rather than shrinking: the dialling and the documenting leave, and what arrives is a smaller number of patients who genuinely need a clinician, with the conversation already attached.
What will this not fix?
A readmission that was always going to happen. Disease progresses, and a check-in does not change the trajectory of an illness that was going to need a hospital.
A patient who cannot afford their medication. Detecting that faster is genuinely useful and does not solve it — the agent surfaces the reason, and somebody with a budget has to act on it.
And an escalation nobody receives. This is the failure worth planning for before anything else: contacting the whole panel generates more escalations than contacting the top decile did. That is the point of it, and it is also a load. An organisation with nobody to receive them has bought a more detailed view of a problem it still cannot fix. Plan the receiving end before the contacting end.
Do readmissions cost your hospital money?
Almost certainly not, unless you are in the United States. This is the most consistently misrepresented claim in our category, including by us until we went and checked, so it is worth setting out precisely.
In the United States, the Centers for Medicare & Medicaid Services runs the Hospital Readmissions Reduction Program, which cuts Medicare payments by up to 3% for hospitals with excess 30-day readmissions in certain conditions. The penalty applies to all Medicare base payments, not just the readmission — so it is a genuine loss, and it is why a US hospital executive takes this meeting.
In the UAE, Saudi Arabia and India, the incentive runs the other way. Inpatient care is paid per episode — by DRG in the Gulf, by fixed package under Ayushman Bharat in India — and none of those systems bundle a readmission into the original episode. So a readmission is admitted, coded and paid as a new one. It earns the hospital money. Any vendor telling a Gulf or Indian CFO to reduce readmissions to protect revenue has the sign backwards.
Australia is the interesting case. It built a national readmission penalty and ran it from 2021 — then switched it off on 1 July 2026, where it now sits shadow-priced while the Commission on Safety and Quality reviews whether it worked, reporting by 30 June 2027. And even while it was live it could not make a readmission cost money: it reduced the original episode by a fraction of the readmission’s price, while the readmission still earned its own funding. At its strongest that is a wash. For a complex patient — which is to say exactly the patient a follow-up programme is for — the hospital stayed ahead.
So outside the US the case has to be built from that system’s own economics, and it is different in each. In Abu Dhabi the honest hook is the opposite of a penalty: the Department of Health’s adjudication rules let a provider and an insurer agree 30-day readmission as a Pay-for-Quality indicator, which pays more for beating your peers. In Australia the currency is the bed day, not the readmission. In India it is not a hospital conversation at all.
We would rather tell you this than sell you a penalty that does not exist in your country. If a vendor quotes you a readmission-penalty argument outside the US, ask them to name the regulation and the date it took effect.
What survives everywhere is the clinical argument, and it is the one worth having: the patient nobody had time to call is a patient nobody is observing. That is true whether or not anyone is fined for it.
Frequently asked
- How do you reduce 30-day readmissions with AI?
- By removing the capacity constraint on post-discharge follow-up. The check-ins that catch a missed medication, a side-effect or early deterioration are well understood but are rationed to high-risk patients because they are thousands of clinician phone calls. An AI agent runs structured contact on every discharge and escalates only the cases that need judgement, so coverage stops depending on staffing.
- Which readmissions are actually preventable?
- Not the ones driven by disease progression — those are unaffected by follow-up. The addressable group tends to share a shape: a small medication failure in the first fortnight that nobody noticed. The prescription was never collected, it was taken incorrectly, a side-effect ended it, or deterioration went unreported. Each is detectable by asking a structured question at the right moment.
- What readmission reduction should we expect from OneDose?
- OneDose does not publish an outcome percentage it cannot attribute to a named deployment, so there is no honest figure to quote here. What can be evaluated before you commit is the mechanism and the arithmetic on your own discharges: how many go uncontacted today, and what structured contact on each would consist of.
- Does reducing readmissions save a hospital money?
- Only in the United States, where CMS runs the Hospital Readmissions Reduction Program and cuts Medicare payments by up to 3% for excess 30-day readmissions. Elsewhere the incentive usually runs the other way. In the UAE, Saudi Arabia and India, inpatient care is paid per episode — by DRG in the Gulf, by fixed package under Ayushman Bharat in India — and a readmission is admitted and paid as a new episode, so it earns the hospital money. Australia built a readmission penalty in 2021 but switched it off on 1 July 2026 pending a review reporting by 30 June 2027, and even while live it reduced the original episode rather than the readmission, so it never imposed a loss. Outside the US the case has to be built from local economics rather than by assuming a US penalty applies.
- Does the AI decide which patients are at risk?
- It scores and surfaces; a clinician decides. Risk scoring determines contact frequency and flags patients for attention. It does not close cases, make clinical determinations, or override a clinician’s judgement about a patient.