The insurance argument that wins
Quality-adjusted life years gained per dollar spent. Reductions in emergency department visits and lifetime disability claims. The advocacy argument has been made for forty years. The actuarial argument requires data the field has not produced.
The argument for insurance coverage of medical foods in inborn errors of metabolism has been made the same way for forty years. The argument is that the medical food is the treatment. Without it, children develop intellectual disability, organ damage, or both. The disability and damage cost the system more than the medical food costs. Therefore the insurer should cover the medical food.
The argument is correct. The argument has not consistently won. State-level coverage mandates exist in roughly half of US states, with substantial variation in scope. Federal coverage through Medicaid is uneven. Private insurance coverage is a patchwork. Families navigate a state-by-state legal and administrative landscape that exists because the underlying argument has been waged through legislation rather than through the kind of evidence insurers actually respond to.
The kind of evidence insurers respond to is cost-effectiveness data framed in their language. Quality-adjusted life years gained per dollar spent. Reductions in emergency department visits, hospitalizations, and long-term disability claims. Comparison of treatment cost against the cost of the alternative trajectory. The argument that wins in coverage discussions is the argument that demonstrates economic outcomes in a form the insurer's actuarial process can ingest.
Why the actuarial argument has been hard to make
The economic argument requires longitudinal outcome data that distinguishes treated from untreated trajectories at the population level. The data has to capture treatment exposure, treatment adherence, clinical outcomes (cognitive, hepatic, renal, cardiac, depending on the condition), and healthcare utilization (emergency visits, hospitalizations, complications, comorbid conditions managed). The same data has to be available across enough patients in enough conditions for the analysis to be statistically meaningful.
The data has not been collected in this form for most rare diseases. Clinical trial data captures treatment outcomes within the trial period but does not extend across decades. Registry data captures clinical outcomes but typically does not include the healthcare utilization data required for cost-effectiveness analysis. Claims data captures utilization but lacks the clinical context that distinguishes adherent from non-adherent patients and treated from suboptimally treated trajectories. The fragments exist; the integrated dataset does not.
The data trust model addresses the fragmentation by holding the integrated longitudinal dataset under patient governance. Treatment data, adherence data, clinical outcomes, and utilization data flow into the same trust under the contributing patient's consent. The trust supports queries that require all four data layers. The cost-effectiveness analysis becomes tractable because the data needed to perform it is in one place rather than in four.
What the analysis enables
The output of the analysis is a comparison that the insurer can act on. For PKU, the comparison is the cost of medical formula across childhood (several thousand dollars per year, varying by age and product) against the cost of the alternative trajectory (special education, lifetime healthcare for cognitive complications of suboptimal control, lost productivity). The current evidence base for the comparison comes from a small number of academic cost-effectiveness papers that consistently find favorable economics for PKU coverage but that do not have the granular cohort data to make case-level coverage decisions.
The granular cohort data is what the data trust produces. A specific 12-year-old with PKU on a specific formula at a specific dose, with documented adherence and metabolic control, has an actuarial profile the insurer can model. The case-level prediction supports case-level coverage decisions. The aggregate of those cases supports policy-level decisions. The data is the same data; the level of analysis is what differs.
The medical food example generalizes to the broader question of rare disease coverage. The Crinecerfont coverage decisions for CAH. The Trikafta coverage decisions for cystic fibrosis variants on the responsive list. The crizanlizumab coverage decisions for sickle cell disease. The casgevy coverage decisions for sickle cell disease. Each of these involves a coverage threshold that depends on outcome data of a kind that the field is producing inconsistently and that the data trust would produce systematically.
The flywheel
The coverage argument and the data infrastructure reinforce each other. As the data infrastructure produces evidence, coverage decisions become more favorable for treatments with documented outcomes. As coverage improves, more patients access treatment, generating more outcome data. As more outcome data accumulates, the evidence base for the next coverage decision strengthens. The cycle is positive feedback when the data infrastructure exists and stagnant when it does not.
The current pattern in rare disease coverage is the stagnant pattern. Coverage decisions are made on the basis of clinical trial data and limited registry data, with the cost-effectiveness analyses produced by academic groups using whatever data is publicly available. The decisions are slow, inconsistent, and litigated state by state.
The accelerated pattern that the data trust enables is data-driven coverage decisions made at speed. A new therapy with strong real-world evidence outcomes data accumulates a coverage profile that supports rapid uptake. A therapy with weaker outcomes data accumulates the evidence base that supports a more measured coverage approach. The decisions are more transparent because the data is more transparent.
The shift from advocacy-driven coverage to evidence-driven coverage is the shift the rare disease community needs. The advocacy approach has worked, slowly, and produced the patchwork that exists. The evidence approach is the route to durable coverage that does not require new state legislation for each new therapy in each new condition. Building the evidence base is the project. The evidence base requires the data infrastructure that has not been built. The infrastructure, once built, produces the evidence the coverage decisions have been waiting for.