A cohort study in JAMA Network Open looked at 202 brain donors from the UNITE Brain Bank — former American football players, recruited between 2008 and 2021 — and compared two things: the dementia diagnosis established by clinicians, and what ended up written on the death certificate.
Fewer than a third had a neurodegenerative disease recorded as the underlying cause of death. The authors estimate the certificate field misses roughly 69% of dementia-related deaths in that population.
The caveat, first
This is a very particular population: men, professional athletes, with repeated head trauma, and brains donated precisely because someone suspected pathology. The number does not transfer to Portugal, nor to a care home in Viseu.
What does transfer is the mechanism. A death certificate is not a research instrument. It is filled in quickly, often by someone who did not follow the person for years, and it tends to record the final event — pneumonia, sepsis, cardiac arrest — rather than the disease that brought the person there.
Why it matters to people who provide care
Because administrative records are the raw material of decisions that later land on institutions.
Ask what dementia costs a country and the answer comes out of databases. Decide how many continuing-care beds are needed, databases. Allocate funding across disease areas, databases. If dementia systematically shows up smaller than it is, everything downstream is calibrated smaller than it should be.
This is not a theory about statistics. It is why a field's budget can be smaller than the field's reality, with nobody able to point at where the count went missing.
We have written about this gap closer to the ground: what the health system measures and what stays invisible. The death certificate is the same problem, at the end of the line.
What an institution can do, and cannot
It cannot correct death certificates. That is a medical act, and whoever performs it is not in the activity room.
It can look after the record while the person is alive, which is the source everything else drinks from:
- Diagnosis in the file, with a date and a source. "Dementia" and nothing else is the start of a lost record.
- Observation instead of impression. "Asked the same question four times in twenty minutes" is a record. "Seems more confused" is not.
- What was tried, and what happened. An activity, a session, a change of routine: what was done, for how long, what was observed. This is what gives a referral letter substance.
None of the three costs anything. All three depend on someone deciding that the record is part of the care and not paperwork after it — the argument we made in the session and the patient's clinical file.
And technology?
A virtual reality system solves none of this, and it is worth saying plainly. What a tool with logging does is more modest: it writes down what happened in a session — duration, scene, what the team observed — and saves the part of the record that gets lost to tiredness at the end of a shift.
That will not correct a death certificate ten years from now. It makes it less likely that Thursday's hour vanishes without a trace.
Source: cohort study in JAMA Network Open on 202 UNITE Brain Bank donors and neurodegenerative disease recording on death certificates, reported by AJMC.