Lower total cost of care, and evidence on who to enroll first.
In a cost-stratified Medicare population, monitoring was associated with a median reduction of $10,932 per patient in a single year. The savings held in every baseline cost quartile, including the lowest, which is the finding regression to the mean cannot explain away.
Cost and utilization, in the published cohort
−$10,932Where this number comes fromPopulation597 high-acuity Medicare beneficiariesSourceTelemedicine and e-Health, 2026VerifiedAugust 2026Observational and single-arm — reported as an association, not causation.
median annual total cost of care per patient, 597 high-acuity Medicare beneficiaries
−40.9%Where this number comes fromPopulation597 high-acuity Medicare beneficiariesSourceTelemedicine and e-Health, 2026VerifiedAugust 2026Observational and single-arm — reported as an association, not causation.
inpatient hospitalizations, 597 high-acuity Medicare beneficiaries
−19.9%Where this number comes fromPopulation597 high-acuity Medicare beneficiariesSourceTelemedicine and e-Health, 2026VerifiedAugust 2026Observational and single-arm — reported as an association, not causation.
emergency department visits, 597 high-acuity Medicare beneficiaries
52.8% → 25.8%
share with at least one hospitalization, 597 high-acuity Medicare beneficiaries
597 high-acuity Medicare beneficiaries, each with annualized pre-enrollment costs above $12,000. All p<0.001. Observational pre–post design, reported as association, not causation.
The savings hold where you would least expect them.
Reductions were statistically significant in all four baseline cost quartiles. In the lowest, patients entering at $12,003 to $17,549, median cost still fell $5,391 (p=0.033). Regression to the mean predicts large apparent savings in the most expensive patients and little in the least expensive; a significant reduction in the lowest quartile is the observation that argument cannot accommodate.
Who to enroll first
Cost, not diagnosis, predicts who benefits.
In multivariable regression, baseline total cost of care, not age, sex or BMI, was the dominant predictor of post-enrollment cost. For an organization designing a deployment, that is the practical finding: cost-based patient selection identifies the population most likely to show an economically meaningful response, and demographic variables add little beyond expenditure history.
The arithmetic, on your numbers
The published year, at your scale.
The two headline results from the 597-beneficiary cost study, multiplied by a panel size you choose. The figures are published; the multiplication is the only thing we have added — and it is labeled for what it is.
≈ $10.9M
implied annual reduction in total cost of care — if every patient matched the study median
−$10,932 median annual total cost of care per patient · 597 high-acuity Medicare beneficiaries · Telemedicine and e-Health, 2026
≈ 270 patients
more ending the year with no hospitalization at all — if the cohort’s shift held
share with at least one hospitalization fell 52.8% → 25.8% · 597 high-acuity Medicare beneficiaries
Read this honestly
The study is observational and single-arm: these are associations, not causation. Its cohort was selected for high baseline cost — the numbers above are what that cohort did in one year, multiplied, not a forecast of yours. Medians do not sum, so the dollar line illustrates scale and nothing more. Whether your population resembles the cohort is exactly the conversation a demo is for.
The controlled comparison
−26%Where this number comes fromPopulation255 monitored patients vs 4,245 controls, a fully risk-bearing ACOSourceHealthSnap ACO program analysis, 2026Verified2026Observational and single-arm — reported as an association, not causation.
annual claims, 255 monitored patients vs 4,245 controls, a fully risk-bearing ACO
$92.50Where this number comes fromPopulation255 monitored patients vs 4,245 controls, a fully risk-bearing ACOSourceHealthSnap ACO program analysis, 2026Verified2026Observational and single-arm — reported as an association, not causation.
PMPM savings, 255 monitored patients vs 4,245 controls, a fully risk-bearing ACO
−13.3%Where this number comes fromPopulation255 monitored patients vs 4,245 controls, a fully risk-bearing ACOSourceHealthSnap ACO program analysis, 2026Verified2026Observational and single-arm — reported as an association, not causation.
ED visits, 255 monitored patients vs 4,245 controls, a fully risk-bearing ACO
In a separate cross-sectional analysis within a fully risk-bearing ACO, 255 monitored patients were compared against 4,245 controls: 26% lower annual claims expense, $92.50 PMPM savings, and a 13.3% reduction in ED visits.
What continuous means, one reading at a time.
The cohort numbers above are aggregates of nights like this one. Here is the path a single reading takes, and who acts on it.
What happens at 3 AM
- 03:14A reading arrivesSystolic 186. Transmitted from a cellular cuff at home — no app, no phone to unlock.
- 03:14Analyzed on arrivalChecked against her thresholds — the program defaults, or the custom ones set for her — and placed on the care team’s queue.
- 08:30A navigator picks it upCalls, confirms symptoms and context, records what she finds against the care plan.
What reaches the practiceA note in the chart, already written and already checked. Most of the time, nothing else.
What you get
Enterprise data-warehouse integration
Remote Care Navigator certification
Customizable education & coordination
Cost-based patient selection, validated by the quartile analysis
HealthSnap prioritizes the sickest patients, promotes continuity post-discharge, and helps keep in-network patients in-network.
For your finance and quality teams
The full evidence base, as submitted to CMS.
A compiled binder of peer-reviewed publications, national platform analyses and scientific presentations, prepared for the Centers for Medicare & Medicaid Services in August 2026 by our Chief Scientific Officer.
Programs available to you
RPM · CCM · APCM · post-discharge readmission reduction, with PMPM and shared-savings arrangements supported.