Research by Max · Feb 21, 2026 · n=18 sources · Confidence: Medium-High
Start with Humana. Best self-service sandbox in the industry — instant credentials, no forms, no approval wait. 7.5M members across Medicare Advantage + commercial. Validate the EOB denial data workflow in days, not weeks.
Then Aetna. 24.9M members, relatively friendly portal, good docs. Together that's ~32M covered lives from two integrations.
For scale without the per-payer grind: Flexpa. $65K/year gets 3 payers via one API. $130K/year gets 400+ payers. They're the biggest consumer of Patient Access APIs in the U.S. — they've already debugged every payer's quirks so UPHELD doesn't have to.
Enrollment: Mark Farrah Associates via Peterson-KFF, Dec 2025. FHIR status: developer portals + healthsouse.com developer reviews.
| # | Insurer | Members (M) | FHIR Status | Dev Portal | Sandbox? | Known Quirks | Priority |
|---|---|---|---|---|---|---|---|
| 8 | Humana | 7.5 | ✅ Live | developers.humana.com | ✅ Self-service, instant | Smaller member base vs top 3; otherwise cleanest devx available | 🟢 #1 — START HERE |
| 3 | CVS Health / Aetna | 24.9 | ✅ Live | developerportal.aetna.com | ✅ Yes (emailed Excel form, fast) | Non-standard OAuth flow; requires emailed security form; initially MA-only | 🔵 #2 — High members |
| 1 | UnitedHealth / Optum | 44.8 | ✅ Live, mature | portal.flex.optum.com | ✅ Yes | Doc/reality mismatch on EOB; two portals can confuse; fragmented sub-portals | 🟡 #3 — Biggest, harder build |
| 2 | Elevance / Anthem | 36.1 | ✅ Live | anthem.com/developers | Likely yes | Multiple brand portals (Anthem/Wellpoint/WellCare); state-by-state variation | 🟡 #4 — High value, complex |
| 6 | Cigna Healthcare | 16.2 | ✅ Live | developer.cigna.com | ✅ Via Help form | Sandbox not fully self-service; auth flow has documented quirks; production needs explicit approval | ⚪ #5 — Medium priority |
| 4 | Centene | 19.0 | ✅ Live | partners.centene.com | Unknown | Primarily Medicaid — lower commercial denial data; different patient profile than UPHELD's core use case | ⚪ #6 — Medicaid-heavy |
| 5 | HCSC (BCBS IL/TX/etc) | 18.7 | ✅ Live | interoperability.hcsc.com | Unknown | Portal appears JS-heavy/fragile; no public sandbox details found in research | ⚪ #7 — Verify portal first |
| 7 | Kaiser Permanente | 12.3 | ✅ Live | developer.kp.org | Yes | Integrated system (insurer + provider) — EOB denial data scope may differ from pure insurers; verify first | ⚪ #8 — Verify EOB scope |
| 9 | GuideWell (Florida Blue) | 6.1 | ✅ Live | developer.bcbsfl.com | ✅ Yes | Florida-only coverage; good portal structure with multiple FHIR products | ⚪ #9 — Regional |
| 10 | BCBS Michigan | 4.9 | Likely ✅ | Unknown (BCBS network) | Unknown | Michigan-only; limited public developer info found in research | ⚪ #10 — Lowest priority |
Verdict: Flexpa is the right call for UPHELD. They're purpose-built for patient-consented FHIR data access, have transparent pricing, and the Essential tier at $65K/year is a real option for an early-stage company proving the concept. 1upHealth is better suited for payers and health systems building their own FHIR infrastructure — not the use case here.
Each payer has their own portal (see table above). Required info: company name, app name, contact email, redirect URI (where to send users after OAuth consent), brief app description. Humana: fully self-service, instant. Aetna: emailed security form, processed within days. Cigna: Help form submission.
Payer issues Client ID + Client Secret. These are scoped to patient data resources: Patient, ExplanationOfBenefit (EOB), Coverage, Practitioner. For Humana sandbox: available immediately after registration at developers.humana.com/apis/registerapp.
Patient navigates to payer's authorization server (Humana sandbox: sandbox-fhir.humana.com/auth/authorize). Patient consents to data sharing. Payer returns authorization code → app exchanges for access token. Standard OAuth 2.0 flow — same as "Login with Google" pattern.
GET https://[payer-fhir-base]/ExplanationOfBenefit?patient=[id] Returns full claims history including denial status, denial reason codes (adjudication.reason), payment amounts, and prior authorization data (as of Jan 2026 mandate). This is the core data UPHELD needs.
Separate production review for most payers: app security review, data use attestation, sometimes legal review. Humana and Aetna are faster than most. Cigna and UHC have historically taken longer. If using Flexpa, this step is handled by Flexpa for each payer in their network — UPHELD only needs one production approval from Flexpa.
Option A: Start with Humana direct (free, ~1–3 days to sandbox, weeks to production). Validate the product. Then go Flexpa for scale. Option B: Go straight to Flexpa ($65–130K/year), skip per-payer work entirely. Depends on runway and conviction level at this stage.
$65K gets 3 payers — enough to prove the concept. $130K is the scale play. If UPHELD is still pre-revenue, $65K is a big commitment. What's the fundraising plan / runway situation?
Humana's 7.5M members are primarily Medicare Advantage. Denial patterns, appeal rates, and patient demographics differ significantly from commercially-insured workers. Which is UPHELD's primary customer? This affects which payers to prioritize and the appeal letter content strategy.
~40% of commercially insured Americans (self-funded employer plans) aren't covered by FHIR mandate. Different integration path required. When does UPHELD need to address this? It's a major expansion of TAM but requires a separate technical approach (not FHIR).
Max can draft the app registration details (app name, description, redirect URI placeholder) so Gin just fills in and submits. Takes ~5 minutes. Would unblock all subsequent technical validation. Want this?
Confidence notes: Member counts (High — Mark Farrah Associates). FHIR live status (High — 9/10 confirmed via portals; BCBS Michigan is Medium/inferred). Flexpa pricing (High — live website). 1upHealth pricing (Low — no public data). Developer experience quirks (Medium — 2021–2024 developer reviews; may not reflect current state).