How Transitions of Care Determine Population Health Outcomes
There is no line item in a Medicare Advantage contract labeled “readmissions.” And yet, ask any Stars executive what keeps them up at night in 2026, and all-cause readmissions is likely at the top of the list. Understanding why reveals how the discharge moment has become the single highest-leverage intervention point in population health and why solving it at scale requires rethinking what a care transition program can be.
A small measure gating a very large check
Medicare Advantage quality bonus payments will exceed $13 billion in 2026, averaging roughly $370–$400 per enrollee per year across bonus-qualifying plans. But that money doesn’t flow on a slope; it flows over a cliff. Contracts rated four stars or higher receive a 5% benchmark bonus (10% in double-bonus counties) and keep a larger share of rebate dollars. Contracts at 3.5 stars receive nothing. There is no partial credit.
Within that rating, the Plan All-Cause Readmissions measure has quietly become one of the heaviest single levers a plan holds.
CMS tripled its weight, from 1x to 3x, beginning with the 2025 Star Ratings. The following year, CMS cut the weights of member-experience measures in half, which mechanically inflated the share of the rating carried by each outcome measure. In two rating cycles, readmissions went from a rounding error to a measure that can decide, on its own, which side of the four-star cliff a contract lands on.
The industry’s problem: plans are underperforming on it.
Plan All-Cause Readmissions ranked among the lowest-performing measures in the most recent Star Ratings; at the same time, the share of enrollees in bonus-qualifying plans fell to its lowest level since 2018.
More contracts are sitting closer to the cliff edge than at any point in years, and the heaviest measure many of them hold is the one trending downward.
Triple-counted economics
What makes readmissions unique is that a single avoidable event hits a plan three ways at once:
- The Stars channel. A weak readmissions rate drags a 3x-weighted measure, threatening the benchmark bonus and rebate percentage that can help a plan enhance its supplemental benefit strategy.
- The medical expense channel. Every readmission is a five-figure claim that is also a pure medical loss with no offsetting revenue, at a moment when elevated inpatient utilization is already compressing margins industry-wide.
- The growth channel. Rebate dollars help fund the $0 premiums, dental coverage, and flexible benefits that win enrollment. No bonus means dropping benefits and losing members in the following enrollment period. A Stars miss becomes a benefit cut, which in turn becomes a membership decline.
No other measure in the program compounds this way.
And unlike most of the Star’s portfolio, readmissions cannot be improved through better data operations. There is no supplemental data feed, chart retrieval program, or documentation sweep that changes whether a member returned to the hospital within 30 days. The measure is an observed outcome. It moves only when care actually changes at the moment of discharge.
The discharge moment doesn’t scale, until it does
Here is the operational trap health plans and risk-bearing providers keep falling into:
Transitions of care are the single most pivotal moment to steer a patient’s recovery trajectory, and they are also the moment traditional care management is least equipped to address.
A human-only approach cannot scale. Skilled transition navigators are scarce and expensive, so organizations ration them, typically to a narrow band of patients flagged as highest risk. Everyone else gets a discharge packet and a phone number. The result is predictable: the “low-complexity” discharges that were never called generate a steady stream of preventable bounce-backs, while navigators burn out working queues that are stale by the time they reach them. Coverage is narrow, timing is slow, and the model breaks the moment volume rises.
Additionally, an AI-only approach cannot be trusted with the moments that matter. Fully automated outreach can reach everyone, but it cannot recognize when a discharged patient’s flat responses signal decompensation, interpret a caregiver’s fear, or make the judgment call that a “routine” transition has quietly become a crisis. In populations where readmissions are concentrated, particularly among patients managing physical and behavioral health conditions together and those with fragile social support, pure automation misses exactly the cases it most needs to catch. And no plan should put an unsupervised algorithm in front of a vulnerable patient.
Automated transitions with a human always in the loop
Any viable solution must rest on a single premise: every discharged patient deserves engagement, and the mode of engagement should match the patient’s complexity, with a human always in the loop.
Leading models run on two coordinated tracks:
Human-first for high-complexity patients. Patients with layered clinical and social needs (such as behavioral health comorbidity, polypharmacy, unstable housing, prior utilization patterns, etc.) are engaged directly by care navigators from the start. Automation works underneath them: real-time discharge awareness, prioritized outreach queues, prepared context, and scheduling support, so navigators spend their time navigating rather than hunting for information. The technology multiplies human capacity instead of replacing it.
AI-first for lower-complexity patients, with human escalation built in. Patients with more straightforward transitions are engaged through intelligent automated outreach that confirms understanding of discharge instructions, verifies medication access, and schedules the follow-up visit. The critical design principle: automation is the front door, but never the last word. Any signal of confusion, deterioration, or unmet need escalates immediately to a human navigator. No patient-facing action happens outside human oversight.
This dual-track architecture produces something neither model can achieve alone: universal coverage at the speed of automation, with human judgment concentrated precisely where it changes outcomes.
Proven against the most complex population first
The most credible way to test a transition model is with complex or hard-to-reach patients, to truly gauge if the strategy will impact those who need it most.
Among high-complexity behavioral health patients discharged from inpatient psychiatric care, those supported by an embedded care navigation program showed a 43% reduction in readmissions compared with those receiving a standard discharge plan. The same program drove 30-day follow-up after hospitalization (FUH) performance to the 95th percentile, on a HEDIS measure widely regarded as one of the most difficult to move, in a population widely regarded as the most difficult to mobilize.
Those results matter for two reasons. First, they were achieved against the steepest gradient. If the model works for patients discharged from inpatient psychiatric care, the mechanics transfer to broader medical and mixed populations, and early results in these populations are showing even more promising performance. Second, they demonstrate that this model moves observed outcome measures, the exact category of Stars measure that cannot be fixed with data operations, documentation, or chart work.
Readmissions became Medicare Advantage’s defining measure because it is where quality, cost, and growth converge on a single number, and because it can only be moved by changing what actually happens in the days after discharge.
Doing that for every patient, every time, and at population scale is not a staffing or a software problem. It is both solved together: automation that never sleeps, and humans who never leave the loop.
Transitions of care are pivotal moments. It’s time the industry treated it that way.
Bamboo Health empowers healthcare organizations to improve physical and behavioral health outcomes through one of the most powerful care collaboration networks with Real-Time Care Intelligence™. To learn more about streamlining your readmission and transitions strategy, visit the Automated Transitions solution page or contact us for a no-obligation discussion.
