Health System Specialty Pharmacy: Build vs. Buy vs. Partner

Thesis

This changes the decision tree. The 2026 options are not “build vs. outsource.” They are four:

De novo solo build — ~$1M pre-inventory capex, 9-12 month URAC accreditation timeline, minimum viability ~15,000 Rx/year. Path of the 2010s. Increasingly hard to execute alone as LDD, payer network, and PBM-steering complexity compounds.
Accelerator partner — payer-owned or payer-linked (Shields [Sycamore-controlled with Evernorth preferred equity], CPS Solutions / Trellis Rx [Optum/UNH], CarePathRx [Evernorth 100%]) — fastest path to LDD access and operational scale, but the health system is accepting mega-payer ownership or financing exposure in the accelerator stack. Structural conflict questions now apply to all three, with Shields materially different from the fully payer-owned assets.
Accelerator partner — payer-independent (Clearway Health only) — Boston Medical Center’s hospital-owned SP accelerator is now the only scaled, payer-independent accelerator in the market. Smaller footprint than the three payer-owned options, but structural independence from PBM capital.
Full outsource / white-bag acceptance — accept margin loss and patient-experience fragmentation in exchange for zero capex and regulatory simplicity.

The common framing “white-bagging mandates are squeezing health-system SP” is empirically weaker than the pushback narrative. IQVIA data shows 98.5% of medical oncology product volume still flows through buy-and-bill. Twelve states have now banned mandatory white/brown bagging. The structural squeeze is real at the PBM-contract and LDD-access layers — not at the dispensing-channel layer.

The decision for a health-system CFO or pharmacy director in 2026 is therefore: does the system have the Rx volume, 340B eligibility, and payer mix to run a solo operation at scale? If not, which accelerator’s structural alignment matches the system’s strategic posture on PBM relationships?

Market Context

Health-system specialty pharmacy count and share are still growing. Hospitals and health systems operate 27% of accredited specialty pharmacies in 2024, up from 15% in 2017. Drug Channels Institute identifies nearly 1,900 accredited specialty dispensing locations in its 2026 report. Total U.S. specialty pharmacy market: $265B dispensed in 2024, +8.0% YoY; specialty drugs are now 39% of total pharmacy dispensing revenue, up from 24% in 2013.

But the top 3 PBM-affiliated SPs still generate two-thirds of specialty Rx revenue. Concentration at the top has not loosened. Non-retail share of oral oncology purchases has risen from 21% in 2017 to 32% in 2024 — the wedge health systems are leaning into.

This is the bifurcation: health-system SP count is growing and oral oncology is moving toward HOPD/clinic channels, but PBM-affiliated SP concentration at the top tier is holding. The growth is in the middle of the market, where health systems compete with each other and with regional independents — not in the top tier dominated by Accredo, CVS Specialty, and Optum Specialty.

De Novo Build: The Numbers

Planning, accreditation, and initial staffing for a 340B specialty pharmacy runs ~$1M before drug inventory. URAC accreditation timeline is 9-12 months from consulting kickoff to final decision. Consulting engagements for standards review run $5K-$10K; the URAC standards doc itself is $295; URAC accreditation fees scale with pharmacy revenue and site count.

Once scaled, the economics are strong. Drug Channels reported that one-quarter of hospital-owned SPs generated >$100M annual revenue in 2022, and >60% dispensed >15,000 prescriptions/year. ~98% of hospital-owned SPs participate in 340B. Scaled operations produce robust margins — the risk is sub-scale.

What’s not in public sources: specific Year 1/Year 2/Year 3 opex bands, IT platform licensing costs (EnterpriseRx, Liberty, OmniSys), and typical payback periods. These numbers are typically behind consulting-engagement paywalls. Health systems evaluating de novo should expect these to require direct diligence rather than benchmark pulls.

Minimum viability threshold: The ~15,000 Rx/year floor implied by Drug Channels data is the practical volume below which solo operations struggle to cover fixed cost. Health systems below this threshold should evaluate accelerator or outsource paths before committing to de novo.

340B Margin Economics

The core economic case for health-system SP rests on the 340B spread — the difference between 340B acquisition cost and commercial reimbursement. The program generated >$40B in profits program-wide in 2019 (most recent comprehensive public analysis); it has grown substantially since.

For branded specialty products with significant HOPD oncology volume, 340B discounts reduce manufacturer net revenue 10-20% on affected units — that 10-20% is the health-system SP’s margin before dispensing cost. Exact per-Rx spread on specific drugs (Humira biosimilars, Keytruda, etc.) is not in public sources; spreads are highly drug-specific and the DCI 2026 Economic Report has the behind-paywall detail.

340B eligibility is the fundamental enabler. Non-340B health systems have structurally weaker build economics. DSH (Disproportionate Share Hospital), CAH (Critical Access Hospital), and FQHC (Federally Qualified Health Center) status are the qualifying conditions. Systems losing 340B eligibility (e.g., through DSH percentage reductions or program enforcement actions) see their build-SP case weaken commensurately.

See Health System SP Economics for the full 340B margin math and accelerator-model financial mechanics.

Accelerator Partners: The 2026 Landscape

Shields Health Solutions (Sycamore / Evernorth)

Shields is the largest independent accelerator by client count: ~80 health system partners, 1,000+ hospitals/clinics, 7M+ patients, 30+ disease states, ~3,000 employees. Claims access to >80% of all LDDs.

The 2025 ownership change is the analysis-defining event. Walgreens originally acquired Shields for ~$970M in 2022 (total outlay $1.37B across 2019-2022). On August 28, 2025, WBA completed divestiture to Sycamore Partners + Stefano Pessina family. Financing included **$3.5B of preferred equity from Evernorth (Cigna/Express Scripts)**.

The structural implication: any health system partnering with Shields in 2026 is now indirectly capitalizing an Evernorth-adjacent asset. Evernorth is Cigna’s PBM, which owns Accredo (the largest specialty pharmacy by Rx volume) and CarePathRx (a competing health-system SP accelerator). Shields’ original value proposition was PBM independence — it did not steer patients to a specific PBM-owned SP. The 2025 capitalization compromises that independence.

For health systems: This should be a material diligence line-item in any 2026 Shields contract discussion. It does not necessarily mean partnering with Shields is the wrong decision — but the structural conflict question now exists where it did not before.

CPS Solutions / Trellis Rx (Optum / UnitedHealth Group)

Trellis Rx was acquired by CPS Solutions in June 2022 (exit for Francisco Partners, reportedly 24x MOIC per Axios Pro). Trellis publishes adherence data — 93% medication adherence vs. 80% industry average — and operates a “fund, build, operate” model similar to Shields, with performance-based fees + profit share under the health system’s brand.

CPS Solutions itself provides pharmacy management services to 800+ hospital clients with 2,500+ pharmacy professionals. On February 1, 2024, UnitedHealth Group / Optum Rx acquired CPS from Frazier Healthcare Partners — corroborated by Drug Channels Institute’s 2026 vertical integration map (Adam Fein, April 2026) and Payer Perspectives. The deal was not publicly announced by either party, consistent with UnitedHealth’s stealth-acquisition pattern across 250+ subsidiary additions in 2024. CPS’s own website still credits Frazier as owner as of April 2026.

For health systems: CPS/Trellis Rx is no longer the “PBM-independent accelerator option” — it is an Optum Rx subsidiary. Health systems evaluating CPS in 2026 should treat it with the same structural-conflict diligence as Shields (Evernorth-adjacent) and CarePathRx (Evernorth-owned). The “counterweight to Shields” positioning that held 2019-2024 is structurally eliminated. Additional diligence: 340B consulting independence when the consultant is a UNH subsidiary is a real RFP question, particularly for 340B-heavy safety-net hospitals.

CarePathRx (Evernorth / Cigna)

CarePathRx is the third major accelerator, founded 2019 by John Figueroa + Nautic Partners, scaled via the December 2020 $400M Chartwell MSO acquisition from UPMC. Evernorth took a 49% minority stake in May-June 2023, then exercised its call option on the remaining 51% in mid-2025 (Oregon HCMO filing July 2025), with the deal closing quietly in February 2026. CarePathRx is now a wholly-owned Evernorth/Cigna subsidiary serving 40+ health systems and ~1,000 hospitals.

CarePathRx differentiates on bundled-stack breadth — it is the only accelerator offering SP + home/ambulatory infusion (Chartwell legacy) + 340B TPA + telepharmacy (PipelineRx) + GPO (ProCure) in a single contract. Home infusion depth is particularly relevant for hospitals that want to capture infusion margin.

For health systems: CarePathRx is now structurally equivalent to Shields in PBM entanglement — both are Evernorth-capitalized. The differentiator is vertical stack: CarePathRx bundles home infusion and 340B TPA where Shields focuses on SP. Cigna/Express Scripts data flow is a 2026 diligence question.

Clearway Health (Boston Medical Center — only remaining payer-independent accelerator)

Clearway Health (founded 2014) is Boston Medical Center’s hospital-owned SP accelerator, operating a “hospital-owned SP build” model for peer health systems. Post-February 2026 (with CarePathRx now payer-owned), Clearway is the only scaled, payer-independent health-system SP accelerator in the U.S. market. Footprint is smaller than the three payer-owned options, but structural independence from PBM capital is a meaningful diligence-criterion advantage for 2026 RFPs.

Fairview Pharmacy Solutions (operator-enablement, not an accelerator)

Fairview Pharmacy Solutions sits in a different lane from Shields, CPS / Trellis, CarePathRx, and Clearway. It is not a capital-heavy build-operate accelerator; it packages Fairview’s own specialty-pharmacy, home-infusion, and pharmacy-benefits operating model into advisory, training, and transformation support for peer health systems. That matters for systems that want payer independence and local ownership but still need operating playbooks, specialty-pharmacy revenue optimization, and practical execution support from an operator rather than a generic consultant.

Walgreens Specialty Pharmacy

Walgreens’ specialty pharmacy network has 265 LDD products as of August 2025. This is the benchmark for the “national SP alternative to health-system SP” path. If a health system outsources specialty dispensing to Walgreens, it captures none of the 340B spread but avoids all capex and operational complexity.

Walgreens’ overall strategic review (the Sycamore divestiture path) creates uncertainty about long-term Walgreens Specialty commitment. Health systems considering full outsource to Walgreens Specialty should factor Walgreens’ own corporate transition into their diligence.

Apexus (positioning clarification)

Apexus is frequently grouped with accelerators, but its business is primarily a 340B GPO, not a build-operate SP partner. Health systems considering Apexus are buying 340B contracting and compliance services, not full SP accelerator infrastructure. This is a different decision than the Shields/CPS/Trellis evaluation.

The White-Bagging Landscape (2026)

The “white-bagging mandates are squeezing health-system SPs” narrative is weaker than the pushback narrative.

Empirical share: IQVIA data shows 98.5% of medical oncology (infused) product volume still flows through buy-and-bill. Payer-reported white-bagging in physician offices has held at 15-20% of covered lives over the past five years. Hospital outpatient remains buy-and-bill dominant.

Most aggressive payer: UnitedHealthcare has commercial white-bagging mandates covering >100 specialty and oncology supportive drugs plus gene therapies.

State-law pushback: As of mid-2025, 12 states have banned mandatory white/brown bagging (AMA State Advocacy Update). Explicit prohibition states include Arkansas, Louisiana, Rhode Island, Texas, Virginia, with additional states beyond these five in the 12-state tally. Louisiana’s law (effective June 1) prohibits insurers from denying payment to participating providers who receive physician-administered drugs from out-of-network pharmacies.

Practical implication for the build-vs-buy decision: In states with white-bagging prohibition, the payer-mandate risk to health-system SP volumes is meaningfully lower. In non-prohibition states with UnitedHealthcare concentration >20% of commercial book, white-bagging exposure is real and should factor into volume-threshold analysis.

See White Bagging and Site-of-Care Economics for the dispensing-channel and state-law detail.

PBM Network and LDD Access

PBM steering: The Big 3 — CVS Caremark, Express Scripts, OptumRx — process 80% of all equivalent prescription claims in 2025. Each actively steers specialty to its affiliated SP (CVS Caremark → CVS Specialty; Express Scripts → Accredo; OptumRx → Optum Specialty). Each PBM’s 2025 formulary excludes 600+ products. Formulary exclusion is the primary mechanism of specialty steering; in-network denial is secondary but significant for health-system SPs.

Health-system SP data on in-network rejection rates by each PBM is not in public sources. The ASHP advocacy community and health-system SP coalition filings are the most likely source for this data; for any specific health system evaluating, the answer requires direct PBM network diligence.

LDD access: Shields claims >80% LDD access — the accelerator path does resolve most LDD friction. Walgreens Specialty Pharmacy has 265 LDD products. NCODA’s 2025 Redefining Oncology Distribution report advocates for the Medically Integrated Pharmacy (MIP) model, which structurally favors health-system and community oncology practices over national PBM SPs.

Practical implication: Going solo de novo, LDD access is still constrained. Going via Shields, access is strong but comes with Evernorth entanglement. Going via CPS/Trellis, LDD breadth is narrower than Shields and the Optum/UNH ownership introduces its own structural-conflict questions. Going via CarePathRx, LDD access depends on accelerator-specific disclosure but Cigna entanglement is the same as Shields. The only structurally payer-independent path among scaled accelerators is Clearway Health — but its scale is meaningfully smaller than the three payer-owned options.

Buyer Decision Framework

Health-system CFOs and pharmacy directors evaluating build-vs-buy-vs-partner in 2026 should answer seven questions in order:

1. Volume threshold

Is projected specialty Rx volume >15,000/year within 3 years of launch? If no, skip de novo and evaluate accelerator or outsource paths. If yes, de novo remains viable but the capex and operational complexity still need evaluation against accelerator economics.

2. 340B eligibility

Is the system securely 340B-eligible (DSH %, CAH status, FQHC) and not at risk of losing eligibility in the evaluation horizon? If eligibility is uncertain, de novo economics are structurally weaker. Evaluate with eligibility-scenario sensitivity.

3. Payer mix and white-bagging exposure

What percentage of commercial book is UnitedHealthcare (most aggressive white-bagging payer)? Is the system in a state with white-bagging prohibition law? High UHC concentration + non-prohibition state = material volume risk regardless of path.

4. PBM relationship posture

Does the system’s strategic posture require payer independence? All three leading accelerators (Shields, CPS/Trellis, CarePathRx) are now payer-owned or payer-capitalized. If independence is a hard criterion, the only options are de novo solo build, Clearway Health (only remaining payer-independent scaled accelerator), or full outsource. If the system is willing to accept payer entanglement in exchange for scale, the next question is which payer (Evernorth or Optum) better aligns with the system’s payer mix and contracting posture.

5. URAC accreditation capacity

Does the system have internal expertise to execute URAC accreditation on a 9-12 month timeline? If not, the accelerator path provides this capability; de novo requires consulting engagements that extend timeline and cost.

6. IT platform integration

EHR/EMR integration, specialty pharmacy IT (EnterpriseRx, Liberty, OmniSys, WellSky CareTend, CoverMyMeds), and revenue cycle systems must align. De novo requires full in-house build; accelerators bring existing platforms (Shields TelemetryRx, etc.).

7. Revenue cycle complexity

Specialty billing is materially different from retail. Systems without established specialty RCM capabilities typically drive to the “partner” decision because the accelerator absorbs this complexity.

Accelerator selection (for questions 5-7)

If the framework points to accelerator partnership, the 2026 accelerator-selection diligence is materially different from the 2022 version:

Factor	Shields (Sycamore/Evernorth)	CPS Solutions / Trellis (Optum/UNH)	CarePathRx (Evernorth)	Clearway Health (BMC)
Scale	~80 health systems, 7M+ patients	800+ hospital clients (CPS); Trellis subset	40+ health systems, ~1,000 hospitals	Smaller; specific scale not publicly quantified
LDD access	>80% (claimed)	Narrower; not publicly quantified	Not publicly quantified	Not publicly quantified
Stack breadth	SP-led; clinical staffing	Pharmacy mgmt + SP + telepharmacy + 340B consulting	SP + home infusion + 340B TPA + GPO (broadest stack)	SP build-operate model
Payer alignment	Evernorth preferred equity ($3.5B, 2025)	Optum Rx wholly-owned (Feb 2024)	Evernorth wholly-owned (Feb 2026)	Payer-independent
Operational maturity	Highest in market	Established; long operating history	Bundled-stack mature (Chartwell legacy)	Smaller but experienced
2026-specific diligence	Evernorth entanglement	340B consulting independence; Optum data flow; Segrave retention	Cigna data flow; retention post-Cigna close	None of the above; smaller scale is the trade-off

Implications

For health systems, the build-vs-buy decision is now also a payer-alignment decision. Partnering with Shields, CPS/Trellis, or CarePathRx can accelerate accreditation, LDD access, payer contracting, and operating maturity, but all three leading scaled options now carry PBM or payer capital ties. Systems that view payer independence as strategic should compare de novo build, Clearway, or operator-enablement models against the slower scale-up and weaker LDD access those paths may imply.

For manufacturers, health-system SP growth changes LDD and 340B strategy. Including health-system SPs can protect prescriber relationships and patient continuity, especially in oncology and IDN-heavy products, but it can also increase 340B exposure and complicate PBM-owned accelerator conflicts. Manufacturer network contracts should specify data rights, 340B treatment, PBM firewall terms, and escalation paths when a health-system SP is both a clinical partner and an economic channel competitor (see Drug Channels specialty pharmacy sources, Evernorth/Shields sources, and Health System SP Economics).

Analyst Notes

The August 2025 Sycamore/Evernorth event is the analysis-defining turn. Anyone evaluating build-vs-buy without pricing this into Shields diligence is using a 2023 framework. Treat the $3.5B Evernorth preferred equity as material for 2026 contract discussions.
Do not overclaim white-bagging is winning. 98.5% of infused oncology is still buy-and-bill; 12 states have banned mandatory white/brown bagging. The squeeze on health-system SP is real at the PBM-contract and LDD-access layers, not at the dispensing-channel layer. Wiki page White Bagging and Site-of-Care Economics covers the channel mechanics in detail.
Apexus is not an accelerator. Frequent mischaracterization. Apexus is a 340B GPO. Reposition accordingly in any vendor evaluation.
All three leading accelerators are now payer-owned. Previous framing of CPS Solutions / Trellis Rx as “the PBM-independent option” is obsolete as of the February 1, 2024 Optum/UNH acquisition (confirmed by Drug Channels Institute’s April 2026 vertical integration map). Combined with Shields (Sycamore + Evernorth preferred, Sept 2025) and CarePathRx (Evernorth 100%, Feb 2026), all three scaled accelerators are mega-payer subsidiaries. Clearway Health (Boston Medical Center) is the only remaining scaled, payer-independent accelerator — structural advantage for 2026 payer-sensitive RFPs, though smaller footprint.
15,000 Rx/year floor is the practical minimum-viability threshold for de novo solo operations, implied by Drug Channels data (>60% of hospital SPs dispense above this level).
~$1M pre-inventory capex + 9-12 month URAC timeline are the most defensible public numbers for de novo build cost and timing. Anything more granular (Year 1/2/3 opex, IT licensing, payback period) requires direct diligence rather than benchmark pulls.
Oncology remains the flagship: Non-retail share of oral oncology purchases: 21% → 32% (2017-2024). Health systems with oncology volume should weight oncology share growth heavily in their evaluation.
Contradiction flag: Shields’ “private standalone” press framing (Sycamore Partners deal release) conflicts with the reality of Evernorth preferred equity. Industry press occasionally conflates “sold to Sycamore” with “fully independent of PBMs.” Both are valid framings depending on what’s being asserted. The wiki page should preserve both threads rather than collapse them.
Unsourced claims flagged: Apexus, McKesson Health Mart Atlas, Cardinal OnPoint specific client counts and pricing disclosures; per-Rx 340B spread on specific named drugs; PBM in-network rejection rates for health-system SPs; latest AHIP state white-bagging law chart (last public version January 2024).

Concepts:

340B Drug Pricing Program
Buy-and-Bill
White Bagging and Site-of-Care Economics
Site-of-Care Optimization
Infusion Services Market
PBM Ecosystem

Related Analyses:

Related Comparisons:

Key Vendors:

Auto-generated cross-references closing audit-surfaced link gaps. Vendors named in the prose above without inline links are listed here so the wiki graph is queryable.

URAC