Not every readmission is preventable

Somewhere this quarter, in a hospital conference room, a vendor is promising to cut hospital wide 30 day readmissions by 30 percent.

Work the math on that. If roughly a quarter of readmissions could have been prevented, a 30 percent reduction across the whole population means catching every single preventable one, and then a few that were never in play to begin with.

Nobody in the room says this out loud. The slide moves on.

What the evidence actually supports

A systematic review of 34 studies put the median proportion of readmissions deemed avoidable at 27 percent, with individual studies ranging anywhere from 5 percent to 79 percent.

A national study of 1,000 general medicine patients readmitted within 30 days across 12 academic medical centers landed in the same neighborhood, finding about a quarter potentially preventable. A review published this year of 481 readmissions at one academic center found that 28 percent were preventable, and the reasons the rest were not are worth reading closely. Disease progression and new, unrelated conditions together accounted for roughly 44 percent of all the readmissions reviewed.

Those patients were not failed by anyone. They got sicker, or something else went wrong. No discharge process, no follow up call, and no risk model was going to change the outcome.

So the ceiling on this work sits somewhere around one in four, and it moves depending on your population. Promises well above that line are marketing.

Why the overpromise is expensive

It is tempting to treat this as harmless enthusiasm. It is not.

A program built on an impossible target fails on schedule. Eighteen months in, the readmission rate has moved two or three points, the board asks what happened, and the honest answer, that the program worked about as well as it could have, sounds like an excuse. Good work gets defunded because it was sold against a number that never existed.

The payment system makes this worse rather than better. CMS penalizes hospitals through the Hospital Readmissions Reduction Program using risk adjusted, all cause 30 day readmission rates across six condition cohorts, with reductions capped at 3 percent of base operating payments for the full fiscal year. All cause means exactly what it says. Nothing in that calculation asks whether a readmission was preventable.

So a hospital is measured on the whole and can only act on the subset. That is not an argument against the program, but it does explain why the overpromise is so tempting, and why it is so costly when it collapses. The gap between what you are judged on and what you can change is precisely where unrealistic targets get sold.

It also costs you the clinical floor. Ask a hospitalist whether every readmission was preventable and watch the reaction. They know which of their patients came back because the discharge was rushed and which came back because the cancer advanced. When a program implies those are the same problem, the people who have to run it stop believing anything else it says.

What separates the two groups

This is the part the honest version of this conversation gets to have.

That national cohort found the factors most strongly associated with preventability were the decision making in the emergency department at the time of readmission, failure to relay important information to outpatient clinicians, and discharging patients too soon. Others included patients being unable to keep appointments after discharge.

The review published this year found a similar pattern among preventable cases, with medical management and medication issues, misalignment of goals of care, and a lack of timely follow up after discharge leading the list.

Read that list again. Information that did not travel. A discharge that happened a day early. An appointment somebody could not get to. A conversation about goals that never happened.

Most of that is not medicine failing. It is logistics, communication and sequence. Even the emergency department finding, which is a clinical judgment rather than a workflow gap, tends to be a judgment made without the one thing that would change it, which is a clear picture of what happened after the last discharge.

All of it is the kind of failure that can be seen coming, if anyone is watching for it.

What this changes about the job

If three quarters of the events were never in play, then the work was never about driving the total rate down. It is about finding the subset.

That is a different problem, and it needs different tools. A model that ranks patients by likelihood of return will hand you plenty of patients who are certain to come back and equally certain to be unreachable. Advanced disease. A new diagnosis nobody saw coming. High risk, low opportunity.

The patients worth your team's next phone call are the ones carrying both risk and a fixable reason for it. A discharge with four medication changes and no scheduled follow up. A patient who did not understand the plan. A handoff that never reached the primary care physician.

Which means a score on its own cannot get you there. You cannot sort for a fixable reason unless the reason is visible, which is the argument we made in Why Clinicians Ignore Risk Scores. A number tells you who might come back. Only the reasoning tells you whether anything can be done about it.

Measure against the right denominator

Here is the practical consequence, and it is the thing most programs get wrong.

If you measure your readmission program against your total readmission rate, you will conclude it failed even when it worked. Prevent a third of the preventable ones in a population with a 15 percent readmission rate, and your headline number moves by roughly one point. That looks like noise. It is not noise. It is most of what was available to you.

Measure the flagged and intervened patients against a comparable baseline instead. Track whether the intervention was actually completed, not just assigned. Then you can say what the program did, in a sentence a CFO can check.

Claims worth retiring

If you are buying in this category, a few things should end the meeting.

A percentage reduction offered without a denominator. A promise to eliminate avoidable readmissions, which is a phrase that only sounds careful. Model accuracy presented as though it were an outcome. And a case study with no comparable baseline, which is not evidence, just a number with a logo on it.

The ceiling on this work is real, and it is lower than the industry likes to admit. Saying so is not a weaker position. It is the only one that survives contact with the second year of a contract.

At Preventra, we build against the subset rather than the total, because the honest number is the one you can still defend after the program has been running for a while. If you want to see what that looks like against your own discharge data, our readmission intelligence overview is the place to start.