The hallway nobody governs.
Specialty pharmacy has two well-managed doors. The front door is initiation: prior authorization, step therapy, formulary design, and the clinical review that happens before a prescription is written. The back door is the acute event: the surgery, the hospitalization, the adverse reaction that forces a coverage decision. Billions of dollars flow through managed processes at both ends.
Between them is a hallway. A member is initiated on semaglutide, adalimumab, pembrolizumab, or buprenorphine. The PA approved access. The refills process. The therapy continues for 12 months, 18 months, 30 months. The cost accrues at $7,000 to $25,000 per month. And no structured process exists to assess whether the therapy is still producing the intended clinical outcome, whether the dose that was escalated six months ago ever generated a response, or whether anyone has looked at this member’s trajectory since the day the prescription was written.
This is not a failure of any existing tool. Claims analytics sees the spend. PA checks eligibility. UM intervenes on individuals. PBMs manage formulary. Each of these instruments was designed for a specific function, and each performs that function. The problem is structural: no instrument was designed to govern the ongoing state of continuation at the population level. The intelligence exists in fragments across four systems. The assembly does not.
That is where Cadence operates. The hallway.
What one cycle reveals.
A single Cadence governance cycle takes 90 days. The input is a de-identified claims extract: four required fields, six optional. Technology flags members meeting configured trigger thresholds across seven required categories. Qualified external clinical reviewers examine each flagged case and assign one of four advisory-only outcomes: Continue, Adjust, Taper, or Switch. No denial authority. No clinical override. The treating provider retains full prescribing authority.
Across three independent populations totaling 65,234 patients, including an independent validation on 30,734 patients in the NIH All of Us Research Program, the governance shape converged:
Sixty percent of reviewed members had clinical trajectories that warranted reassessment. Forty percent were confirmed appropriate by a qualified reviewer, member by member. Half of all advisory recommendations persisted at the second cycle without compulsion, without override, without denial authority. The treating provider independently arrived at a similar conclusion because the signal was clinically sound.
This is what a single cycle produces. An artifact your board can hold. A certificate your stop-loss carrier can verify. And the first measured governance baseline your organization has ever had for continuation therapy.
That baseline is where Sustain begins.
What happens when the governance stays on.
A single cycle is a diagnostic. It shows you the shape of the gap. Sustain is what happens when the governance layer remains in place and each cycle builds on the last.
Calibration
The first cycle uses population-level trigger thresholds drawn from the Cadence Governance Standard. The second cycle starts from a calibrated baseline. Triggers are refined to your population’s actual clinical patterns. A self-funded employer with a high-cost biologic cohort does not need the same threshold sensitivity as a Medicaid plan managing GLP-1 continuation. After one cycle, the configuration fingerprint reflects your population, not a reference model. Every subsequent cycle sharpens the signal.
Compounding
The second governance cycle across the original cohort produced a 52% RIR and approximately $7.1M in additional documented governance signal. Cumulative two-cycle value: $15.7M. The 8-point RIR decline from Cycle 1 (60%) to Cycle 2 (52%) is the governance working. The easiest cases to influence were influenced in the first pass. The remaining population is harder, and the governance finds less low-hanging signal. That attenuation is a sign of a functioning system, not a declining one.
Persistence data sharpens with each pass. The first-cycle GPR (50%) tells you that half of advisory signals persisted. The second-cycle GPR will tell you which therapy classes, which trigger types, and which provider behaviors respond most to governance signals. By cycle three, the governance intelligence is predictive: you know which members are most likely to benefit from reassessment before the cycle begins. And a plan whose RIR has declined to 15% by cycle five holds an 85% DAR: documented confirmation that the vast majority of its continuation population is clinically appropriate. The governance found less to change because there was less to find.
Benchmarking
After one cycle, your RIR is a number. After two cycles, it is a trend. After three, it is a benchmark with a trajectory. You can compare your governance posture cycle-over-cycle, and as the Cadence client base grows, you can compare your posture against anonymized cross-plan benchmarks. The standard that measures you is the same standard that measures everyone. The benchmark is structural, not self-reported.
One cycle shows you the gap. Ongoing cycles govern it. The difference between a governance assessment and a governance layer is whether it stays on.
Three phases. One relationship.
A single 90-day cycle. The first artifact. The first measured RIR and DAR. The Governance Certificate for your next stop-loss renewal. Your organization sees the shape of its continuation governance gap for the first time. Pricing: fixed PMPM or standalone Facilitated Assessment ($15–25K). The buyer’s risk is one cycle. The output is permanent.
Continuous 90-day cycles. Triggers calibrated to your population. RIR benchmarked against your own prior data. GPR tracking tells you which signals persisted and which therapy classes respond to governance. Your stop-loss carrier sees a second certificate, then a third, each documenting a governance posture that compounds. Pricing: PMPM, scaled to cohort size. The governance layer is now a permanent part of your benefits architecture.
After two or more completed cycles with verified outcomes data, the economics can shift from fixed PMPM to performance-aligned pricing tied to documented governance value. The plan pays less at entry. Cadence participates in the value the governance surfaces. The incentive structure aligns: the governance entity earns more when the governance works harder. The margin is not close. At reference parameters, the governance cycle needs to find trajectory changes in just 3.1% of reviewed cases to justify its cost. Two independent cohorts found 58–60%.
The phases are sequential but not mandatory. An organization can remain at Phase 1 indefinitely, running periodic governance diagnostics. Phase 2 is for organizations that want continuous oversight. Phase 3 is for mature engagements where both parties have enough data to share in the outcome. Each phase earns the next. Nothing is assumed.
Why the layer scales.
Most healthcare services require infrastructure that grows linearly with the population served. More patients require more clinicians, more facilities, more provider contracts, more technology integrations. Cadence Sustain does not.
No provider network. Cadence does not contract with treating providers. It does not negotiate rates, manage panels, or depend on provider willingness to participate. The governance operates at the plan level, upstream of the provider relationship. Advisory signals reach the provider through existing channels. The treating physician retains full authority. This eliminates the single largest source of friction and cost in healthcare services: network construction.
No technology integration. The input is a CSV. Four fields. The same format your claims analytics vendors already receive. No EHR integration. No API. No implementation timeline. No IT committee. A governance cycle can begin the week after a signed agreement.
No denial authority. Every determination is advisory. Cadence never denies, restricts, or overrides a prescription. This is the structural decision that makes everything else possible. An advisory-only governance layer does not trigger the duty to act. It does not create the fiduciary cascade that prevents internal teams from building this function. It does not require the compliance infrastructure that makes UM programs expensive and slow. The governance layer is lightweight because it chose to inform rather than intervene.
The human judgment is the product. Technology handles flagging, data assembly, and queue prioritization. The reviewer examines an assembled clinical picture and makes a single governance determination. Measured time: approximately four minutes per case. At 120 cases per reviewer per day, a single qualified PharmD can govern a 25,000-member cohort within one 90-day cycle. The marginal cost of adding the next 10,000 members is one additional part-time reviewer, not a new department.
An algorithm can flag a member. It cannot sign a governance determination. The signature is the product. Everything else is infrastructure that makes the signature possible.
What you hold after year one.
An organization that completes three governance cycles under the Cadence Governance Standard holds the following, none of which existed before the first cycle:
Three governance artifacts. Each containing a configuration fingerprint, outcome distribution, measured RIR and DAR, and an immutable audit trail documenting every reviewed case. Sealed at cycle close. Versioned. Comparable across cycles.
Three Governance Certificates. Each attesting to a completed governance cycle under the published standard. Your stop-loss carrier has seen documented continuation oversight from your organization three consecutive quarters. No other plan sponsor they underwrite can produce this.
A governance trajectory. RIR trending cycle-over-cycle. GPR data showing which signals persisted and which therapy classes responded. A calibrated trigger configuration tuned to your population. The beginning of a predictive governance model specific to your member base.
A fiduciary record. Documented evidence that structured oversight of high-cost continuation therapy was exercised quarterly, under an external published standard, by qualified clinical reviewers, with full audit trail. This is the answer to the question your board has not yet asked but will: what governance did we exercise over the $40M in specialty continuation spend that flowed through our plan last year?
The question is coming. ERISA requires prudent stewardship. The Consolidated Appropriations Act of 2026 increases fiduciary scrutiny of pharmacy benefit decisions. CMS’s BALANCE Model is expanding the continuation population. Stop-loss carriers are tightening underwriting. The organizations that can answer the governance question are the ones that started before it was asked.