Approval is not access
A regulator's clearance says a device is safe and effective enough to sell. It says nothing about whether anyone will pay for it. Reimbursement and access is a separate gauntlet: a payer must issue a coverage decision, a billing code must exist, and a payment amount must be set — often years after approval and via evidence that regulators never asked for. Many cleared devices reach almost no one because this second system, not the first, is where they stall.
The health-economics of an implant
Payers reason in cost-effectiveness. The standard summary is the incremental cost-effectiveness ratio (ICER): the extra cost of the new therapy divided by the extra health it buys, measured in quality-adjusted life-years (QALYs) — years of life weighted by a health-utility between 0 and 1. A therapy is judged against a willingness-to-pay threshold (cost per QALY).
\text{ICER} \;=\; \frac{C_{1}-C_{0}}{E_{1}-E_{0}} \;=\; \frac{\Delta\text{Cost}}{\Delta\text{QALY}}ICER compares a new therapy (1) to standard care (0); a lower cost per QALY gained is more likely to be funded.
A single number payers use to judge value: how much extra a new therapy costs for each extra unit of health (a QALY) it buys, compared to standard care. A lower cost-per-QALY is easier to fund; a therapy that costs a fortune for tiny health gains lands on the wrong side of the threshold. This drives reimbursement.
- \text{ICER}
- Incremental cost-effectiveness ratio — extra cost per extra QALY.
- C_1 - C_0
- The added cost of the new therapy over standard care.
- E_1 - E_0
- The added health benefit, measured in QALYs.
- \Delta\text{QALY}
- Quality-adjusted life-years gained.
A device that costs 50,000 more per patient and adds 2 QALYs has an ICER of 25,000 per QALY — comfortably under many funders' thresholds.
BCI stresses this machinery in a peculiar way: a large one-time cost (device plus surgery) delivers benefit spread over many future years, so the numbers hinge on durability and on discounting the future. A QALY next decade is valued less than one today, and a stream of benefit is summed with a discount rate r over a horizon H. If the implant lasts three years instead of fifteen, the same upfront cost buys a fifth of the discounted benefit — which is why chronic longevity (from the hardware track) is not just an engineering metric but an economic one.
\text{QALY}_{\text{total}} \;=\; \sum_{t=1}^{H} \frac{u_t}{(1+r)^{\,t}}Discounted QALYs: u_t is the health-utility in year t, r the discount rate, H the horizon. Short device lifetime collapses the sum — durability is an economic variable.
Health gained in the future counts a little less than health gained today, so we "discount" each year's benefit. Add up the yearly health utilities, each shrunk by a compounding discount factor. Because the sum stops at the horizon H, a device that wears out early simply has fewer terms to add — making durability an economic, not just a technical, virtue.
- \text{QALY}_{\text{total}}
- The total discounted health benefit over the horizon.
- u_t
- The health utility in year t — roughly, quality of life that year.
- r
- The discount rate — how much future health is marked down.
- H
- The horizon — how many years the benefit lasts.
At a 3% discount rate, a QALY earned ten years out is worth about three-quarters of one earned now; a device lasting only three years accrues just three discounted terms.
Evidence keeps growing after approval
Approval is a checkpoint, not a finish line. Post-market surveillance, patient registries and real-world evidence track durability, rare adverse events, and performance outside the clean trial cohort — the things a small pivotal trial cannot see. A structural weakness of the field today is the absence of standardized outcomes: labs report different metrics on different tasks, so results resist pooling and meta-analysis. Regulatory science — shared benchmarks, common data elements, agreed endpoints — is as much a research frontier as any decoder.
Who gets the future first
Equity of access is not a footnote. Trials cluster at a few elite academic centres, so enrolment skews toward those who live nearby, speak the study's language, and have social support to attend for years — quietly narrowing who the evidence even represents. If the first approved BCIs are priced as premium implants, they risk becoming a technology of the already-advantaged, while the global majority of people with paralysis or ALS remain out of reach. Designing for cost, for languages beyond English, and for lower-resource settings is an ethical and scientific requirement, not charity.
The open problems of translation
Several genuinely unsolved problems stand between today's demonstrations and a routine therapy. How do you run a rigorous randomized trial for a deeply personalized, adaptive device whose decoder is retrained per user and updated over time? How do regulators oversee a decoder that learns — software that changes after approval? Frameworks for Software as a Medical Device and predetermined change-control plans are early attempts, and BCI, where the algorithm adapts continuously, is the hardest test case. Add long-term durability, manufacturing and surgical scaling, and a workforce of trained implanters, and the honest picture emerges.