teleo-codex/agents/vida/musings/research-ma-senior-care-2026-03-10.md

---
status: seed
type: musing
stage: developing
created: 2026-03-10
last_updated: 2026-03-10
tags: [medicare-advantage, senior-care, international-comparison, research-session]
---

# Research Session: Medicare Advantage, Senior Care & International Benchmarks

## What I Found

### Track 1: Medicare Advantage — The Full Picture

The MA story is more structurally complex than our KB currently captures. Three key findings:

**1. MA growth is policy-created, not market-driven.** The 1997-2003 BBA→MMA cycle proves this definitively. When payments were constrained (BBA), plans exited and enrollment crashed 30%. When payments were boosted above FFS (MMA), enrollment exploded. The current 54% penetration is built on a foundation of deliberate overpayment, not demonstrated efficiency. The ideological shift from "cost containment" to "market accommodation" under Republican control in 2003 was the true inflection.

**2. The overpayment is dual-mechanism and self-reinforcing.** MedPAC's $84B/year figure breaks into coding intensity ($40B) and favorable selection ($44B). USC Schaeffer's research reveals the competitive dynamics: aggressive upcoding → better benefits → more enrollees → more revenue → more upcoding. Plans that code accurately are at a structural competitive disadvantage. This is a market failure embedded in the payment design.

**3. Beneficiary savings create political lock-in.** MA saves enrollees 18-24% on OOP costs (~$140/month). With 33M+ beneficiaries, reform is politically radioactive. The concentrated-benefit/diffuse-cost dynamic means MA reform faces the same political economy barrier as every entitlement — even when the fiscal case is overwhelming ($1.2T overpayment over a decade).

**2027 as structural inflection:** V28 completion + chart review exclusion + flat rates = first sustained compression since BBA 1997. The question: does this trigger plan exits (1997 repeat) or differentiation (purpose-built models survive, acquisition-based fail)?

### Track 2: Senior Care Infrastructure

**Home health is the structural winner** — 52% lower costs for heart failure, 94% patient preference, $265B McKinsey shift projection. But the enabling infrastructure (RPM, home health workforce) is still scaling.

**PACE is the existence proof AND the puzzle.** 50 years of operation, proven nursing home avoidance, ~90K enrollees out of 67M eligible (0.13%). If the attractor state is real, why hasn't the most fully integrated capitated model scaled? Capital requirements, awareness, geographic concentration, and regulatory complexity. But for-profit entry in 2025 and 12% growth may signal inflection.

CLAIM CANDIDATE: PACE's 50-year failure to scale despite proven outcomes is the strongest evidence that the healthcare attractor state faces structural barriers beyond payment model design.

**The caregiver crisis is healthcare's hidden subsidy.** 63M unpaid caregivers providing $870B/year in care. This is 16% of the total health economy, invisible to every financial model. The 45% increase over a decade (53M→63M) signals the gap between care needs and institutional capacity is widening, not narrowing.

**Medicare solvency timeline collapsed.** Trust fund exhaustion moved from 2055 to 2040 in less than a year (Big Beautiful Bill). Combined with MA overpayments and demographic pressure (67M 65+ by 2030), the fiscal collision course makes structural reform a matter of when, not whether.

### Track 3: International Comparison

**The US paradox:** 2nd in care process, LAST in outcomes (Commonwealth Fund Mirror Mirror 2024). This is the strongest international evidence for Belief 2 — clinical excellence alone does not produce population health. The problem is structural (access, equity, social determinants), not clinical.

**Costa Rica as strongest counterfactual.** EBAIS model: near-US life expectancy at 1/10 spending. Community-based primary care teams with geographic empanelment — structurally identical to PACE but at national scale. Exemplars in Global Health explicitly argues this is replicable organizational design, not cultural magic.

**Japan's LTCI: the road not taken.** Mandatory universal long-term care insurance since 2000. 25 years of operation proves it's viable and durable. Coverage: 17% of 65+ population receives benefits. The US equivalent would serve ~11.4M people. Currently: PACE (90K) + institutional Medicaid (few million) + 63M unpaid family caregivers.

**Singapore's 3M: the philosophical alternative.** Individual responsibility (mandatory savings) + universal coverage (MediShield Life) + safety net (MediFund). 4.5% of GDP vs. US 18% with comparable outcomes. Proves individual responsibility and universal coverage are not mutually exclusive — challenging the US political binary.

**NHS as cautionary tale.** 3rd overall in Mirror Mirror despite 263% increase in respiratory waiting lists. Proves universal coverage is necessary but not sufficient — underfunding degrades specialty access even in well-designed systems.

## Key Surprises

1. **Favorable selection is almost as large as upcoding.** $44B vs $40B. The narrative focuses on coding fraud, but the bigger story is that MA structurally attracts healthier members. This is by design (prior authorization, narrow networks), not criminal.

2. **PACE costs MORE for Medicaid.** It restructures costs (less acute, more chronic) rather than reducing them. The "prevention saves money" narrative is more complicated than our attractor state thesis assumes.

3. **The US ranks 2nd in care process.** The clinical quality is near-best in the world. The failure is entirely structural — access, equity, social determinants. This is the strongest validation of Belief 2 from international data.

4. **The 2055→2040 solvency collapse.** One tax bill erased 12 years of Medicare solvency. The fiscal fragility is extreme.

5. **The UHC-Optum 17%/61% self-dealing premium.** Vertical integration isn't about efficiency — it's about market power extraction.

## Gaps to Fill

- **GLP-1 interaction with MA economics.** How does GLP-1 prescribing under MA capitation work? Does capitation incentivize or discourage GLP-1 use?
- **Racial disparities in MA.** KFF data shows geographic concentration in majority-minority areas (SNPs in PR, MS, AR). How do MA quality metrics vary by race?
- **Hospital-at-home waiver.** CMS waiver program allowing acute hospital care at home. How is it interacting with the facility-to-home shift?
- **Medicaid expansion interaction.** How does Medicaid expansion in some states vs. not affect the MA landscape and dual-eligible care?
- **Australia and Netherlands deep dives.** They rank #1 and #2 — what's their structural mechanism? Neither is single-payer.

## Belief Updates

**Belief 2 (health outcomes 80-90% non-clinical): STRONGER.** Commonwealth Fund data showing US 2nd in care process, last in outcomes is the strongest international validation yet. If clinical quality were the binding constraint, the US would have the best outcomes.

**Belief 3 (structural misalignment): STRONGER and MORE SPECIFIC.** The MA research reveals that misalignment isn't just fee-for-service vs. value-based. MA is value-based in form but misaligned in practice through coding intensity, favorable selection, and vertical integration self-dealing. The misalignment is deeper than payment model — it's embedded in risk adjustment, competitive dynamics, and political economy.

**Belief 4 (atoms-to-bits boundary): COMPLICATED.** The home health data supports the atoms-to-bits thesis (RPM enabling care at home), but PACE's 50-year failure to scale despite being the most atoms-to-bits-integrated model suggests technology alone doesn't overcome structural barriers. Capital requirements, regulatory complexity, and awareness matter as much as the technology.

## Follow-Up Directions

1. **Deep dive on V28 + chart review exclusion impact modeling.** Which MA plans are most exposed? Can we predict market structure changes?
2. **PACE + for-profit entry analysis.** Is InnovAge or other for-profit PACE operators demonstrating different scaling economics?
3. **Costa Rica EBAIS replication attempts.** Have other countries tried to replicate the EBAIS model? What happened?
4. **Japan LTCI 25-year retrospective.** How have costs evolved? Is it still fiscally sustainable at 28.4% elderly?
5. **Australia/Netherlands system deep dives.** What makes #1 and #2 work?

SOURCE: 18 archives created across all three tracks