Every physician who cares for a patient with dementia starts from scratch. The PCP orders the labs. Geriatrics does the cognitive testing. Neurology gets a referral — and then re-walks the entire workup that's already buried in notes from three months ago. There's no shared checklist. No unified view. No easy way to see, at a glance, where the patient actually is in the diagnostic process.
Early intervention for dementia is important, and only matters if the diagnosis comes in time. Every redundant referral and duplicated assessment is time the patient doesn't have. What’s the solution?
Why the Dementia Workup Is Broken—And How to Fix It
I rotated through geriatrics clinic several times during residency, doing assessments for memory loss, cognitive impairment, falls, incontinence, you name it. As someone heading into critical care, it was not exactly my most thrilling rotation. Nothing gets my heart pumping like a MoCA assessment!
Even without the thrill, my “inefficiency” radar was still running. My conclusion was simple: the dementia workup can be streamlined x1000.
Here’s the current process for working up dementia, based on what I’ve seen.
A family member or PCP suspects significant cognitive impairment (impaired IADLs, memory issues (working memory, short term memory, long term memory)). They request a work up.
The dementia work up is started, but it can get fragmented across referrals and specialists. For example, the PCP may order the typical labs right off the bat (CBC, CMP, TSH, b12, RPR) and refer to neurology or geriatrics.
The specialist (let’s say geriatrics, in this case) does the specialized testing: Mini-Cog, MMSE, MoCA, functional assessments, PHQ-9. Collateral from family members and friends is also imperative, given the patient may not have insight into their cognitive deficits (if any).
Based on these cognitive tests, if dementia is suspected, an MRIb may be ordered. If there’s concern for Alzheimer’s or front-temporal dementia, for example, a neurology referral is made.
The patient sees the neurologist, and the neurologist has to re-walk much of the work up that’s already been done, which may be buried in notes, labs from months ago, and fragmented EHR data.
What I’m getting at is that there’s no easy way for the physician to see, clearly and in an organized way, what work up has already been done for dementia. There’s no obvious place for a “checklist” or a “bundle.” Personally, I think that could speed up the work up and streamline care between specialists.
Root Cause Analysis: 5 Whys
The 5 Whys process in root cause analysis involves repeatedly asking "Why?" five times to drill down into the root cause of a problem by exploring the cause-and-effect relationships underlying the issue.
The problem: The dementia workup is fragmented across specialists, with no standardized bundle or shared checklist, leaving each new clinician to re-piece together what's already been done.
Why?: Multiple specialists (PCP, geriatrics, neurology) are involved sequentially, and each documents independently within their own workflow.
Why?: There is no standardized dementia care bundle or shared checklist that travels with the patient across encounters and specialties.
Why?: No single specialty "owns" the dementia workup pathway — it crosses primary care, geriatrics, and neurology, so coordination defaults to whoever is seeing the patient that day.
Why?: EHR systems are designed for individual-encounter documentation, not for coordinating multi-specialty, sequential diagnostic workups across departments.
Why (root cause)?: Healthcare IT infrastructure is built primarily around billing and documentation rather than clinical decision support and care coordination.
Impact Analysis
Impact analysis is the assessment of the potential consequences and effects that changes in one part of a system may have on other parts of the system or the whole.
Patient: Delayed diagnosis means delayed treatment (which matters when it comes to dementia). Patients cycle through multiple appointments over months before a diagnosis is confirmed. Families carry the burden of coordinating between specialists, repeating history at every visit, and navigating a workup that has no clear endpoint or timeline. For many, by the time a diagnosis is made, the disease has already progressed to a point where earlier interventions—lifestyle modifications, cholinesterase inhibitors, caregiver planning—would have had more impact.
Clinician or Provider: Every specialist who enters the picture starts from scratch. The neurologist re-reviews labs already ordered by the PCP. The geriatrician re-administers cognitive testing the PCP may have already attempted. Time is wasted, documentation is duplicated, and clinical momentum stalls. There's also a cognitive burden when physicians spend energy hunting through fragmented notes rather than synthesizing what's already known and moving forward.
System: Fragmented workups drive unnecessary utilization. Duplicate labs, redundant appointments, and avoidable referrals add cost without adding value. At scale, the absence of a standardized dementia bundle (or care journey) means the system has no way to measure whether the workup is complete, timely, or equitable, which makes quality improvement nearly impossible.
Solution
The solution I envision is a physician-facing “care journey” for non-specific diagnoses that require additional workup to land on a specific diagnosis. For example, “memory loss” is non-specific. A concussion can cause memory loss. A wild night of binge drinking can cause memory loss. Dementia can cause memory loss. Medications can cause memory loss.
So imagine that when you add “memory loss” as a diagnosis to a visit, the patient is automatically enrolled in a pathway that tracks the “next steps” and organizes the diagnostic workup.
[ ] Review medications
[ ] Significant physical exam findings (bradykinesia, cogwheel, tremors)
[ ] IADLs reviewed
[ ] CBC, BMP, B12, TSH, RPR
[ ] MoCA/MME/MiniCog
[ ] Brain imaging
[ ] Medications to consider (acetylcholinesterase inhibitors, memantine)
The challenge, of course, is that this workup rarely happens in one place. Different specialists document in different notes, and the checklist only works if it's populated. That's where AI fits in. Even with fragmentation, AI can pull from unstructured data and populate the care journey automatically. The treating physician can click the patient’s “Memory Loss Care Journey” and quickly see what’s been done so far. They may see that the physical exam and labs were unremarkable, but the patient scored a 15 on the MoCA and can no longer pay bills. From there, the treating physician may order an MRI and refer to neurology. Then the neurologist can immediately see what’s already been completed and decide on next steps.
There’s already something similar to this for cancer treatment. In Epic, if a patient is on active chemotherapy, I can hover over their “Treatment Journey” and quickly see which cycle they’re on, any adverse reactions, how many treatments are left, and which labs are obtained with each treatment. It’s helpful and organized.
You can apply the same idea to other non-specific diagnoses, like "chest pain," "shortness of breath," and "hematuria." The dementia workup just happens to be one of the most fragmented, most time-sensitive, and most consequential places to start.
