The arithmetic inside a large health system is brutally simple, and it is the reason so many families drive two hours for a thirty minute appointment. Sutter Health cares for more than 3.6 million patients in Northern California. It has more than 100 general neurologists. It has five cognitive and behavioral neurologists, and they sit in two specialty memory centers. When lecanemab and donanemab arrived, each one demanding amyloid confirmation, genotype counselling, infusion logistics and repeated MRI surveillance, the conventional answer was to funnel every candidate toward those two addresses. Sutter tried the opposite: keep the patient with their own neurologist, and move the expertise.
The mechanism is a weekly virtual session called the Advanced Therapeutics for Alzheimer's Disease Review Board, staffed by cognitive neurologists, a neuroradiologist, a pharmacist and a registered nurse program manager. A general neurologist submits a case; the board reads the imaging, weighs the eligibility, and sends back a recommendation with the reasoning attached. Referring clinicians can sit in on the discussion, and can re-refer a patient when something goes wrong mid-treatment. The design was first described publicly in a 2025 conference abstract from the same group. The results were published on October 9, 2026 in Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring.
The numbers cover November 2023 through December 2025. The board received 574 referrals. The analysis covers 447 fully reviewed cases submitted by 51 general neurologists, of whom 344, or 77 percent, were judged eligible for treatment. Annual referral volume rose 148 percent, from 163 cases in 2024 to 404 in 2025, which tells you the community neurologists kept coming back. Among patients who went on to treatment, any radiographic ARIA (amyloid related imaging abnormality, the brain swelling and small bleeds these drugs can cause) occurred in 17.4 percent on lecanemab and 28.8 percent on donanemab. Symptomatic ARIA, the kind a person actually feels, occurred in 2.6 percent and 1.7 percent respectively. The authors report these rates were lower than in the pivotal phase 3 trials and consistent with real-world series from specialty memory clinics.
That last comparison deserves care rather than applause. This is a retrospective report from one health system, with no control group and no randomisation, so it cannot tell you whether the board itself lowered anyone's ARIA risk; a population of locally referred patients may simply differ from a trial population. Other real-world series land in similar territory. A retrospective review of 172 amyloid-positive patients published in the same journal on August 26, 2026 found ARIA in 11.2 percent on lecanemab and 13.3 percent on donanemab. Sutter's own earlier cohort of 210 lecanemab patients treated between July 2023 and June 2024 reported ARIA in 15.7 percent, discontinuation in 14 percent, and five deaths, none attributed to the drug.
For a family, the practical meaning is geographic rather than pharmacological. Nothing here says the drugs work better in a community clinic. What it says is that a system can build a structure in which the specialist reads the MRI and argues the eligibility case without the patient being in the specialist's building, and that the resulting safety signals in this cohort looked like those reported from memory centres. Armen Moughamian, the lead author and chief of memory for Sutter's Neuroscience Service Line, framed the problem in the health system's announcement by noting that putting these therapies into practice safely takes "much more than writing a prescription". The reported figures describe who got reviewed, who was deemed eligible, and what the scans showed. They do not report cognitive outcomes.
The open question is whether the model travels. Sutter's Neuroscience Service Line funded the infrastructure, standardised the workflow and pushed guidance across the system, which is a particular kind of institutional will that not every network has. A smaller group without a neuroradiologist on call, or without a nurse program manager tracking infusion schedules and MRI intervals, is not running the same thing merely because it has a video link.