A Dementia Diagnosis Cut Life-Sustaining Care Recommendations Nearly in Half. A Directive Only Partly Closed the Gap.

Nearly 6,000 U.S. adults read the same hospital scenario with one detail changed: whether the patient had dementia. When the patient had dementia, respondents recommended life-sustaining treatment in 15.6 percent of cases. Without dementia, the figure was 38.9 percent. The gap held even when a signed advance directive asked for aggressive care, according to a study published October 7 in JAMA Network Open.

For anyone who has filled out a directive on the assumption that the document decides the outcome regardless of diagnosis, the finding is a flag. It did not.

The study, led by Lauren Hersch Nicholas, a geriatrics professor at the University of Colorado Anschutz School of Medicine, with co-authors from the University of Pennsylvania, the University of Michigan, and Johns Hopkins, ran 11,865 randomized vignettes past 5,965 respondents. Each described a seriously ill, hospitalized older adult and varied three things: the presence of dementia, the content of an advance directive, and what the attending physician recommended. The underlying survey ran in December 2022, nearly four years before the paper's publication.

What the paperwork changed, and what it didn't

An advance directive moved the numbers for every patient, dementia or not. Among patients with dementia, a directive requesting life-sustaining treatment raised the recommendation rate from 15.6 percent to 41.0 percent, more than double. A directive requesting comfort care alone dropped the rate to 7.6 percent, the lowest of any scenario tested.

Patients without dementia saw the same kind of directive do more work. A life-sustaining directive took their recommendation rate from 38.9 percent to 66.3 percent. Line up both groups under identical paperwork and the dementia patient still trailed by roughly 25 percentage points.

Why a document doesn't erase the gap

Nicholas and her co-authors point to assumptions, not neglect. The researchers write that some respondents may be acting on a belief that a dementia diagnosis forecloses a good quality of life, a belief the paper does not test directly but flags as a likely driver of the gap.

“People's assumptions about what life is like with dementia appear to play an important role in how they think about treatment decisions.”

The authors are explicit about what the study cannot show. A written vignette carries none of the emotional strain of a real bedside decision, and the scenarios did not specify survival odds, dementia severity, or the patient's quality of life before hospitalization. A respondent choosing on a screen is not the same as a family member choosing in a hallway outside an ICU.

That leaves two readings of the same numbers. One is that advance directives are failing patients with dementia. The other is that directives are working, just not enough to offset a separate bias that shows up regardless of what the paperwork says. The data support the second reading better than the first: a directive requesting life support roughly tripled the recommendation rate for dementia patients, from 15.6 to 41.0 percent. It still didn't approach the 66.3 percent rate seen in patients without dementia under that same directive.

Smaller studies of surrogate decision-making have raised similar concerns about proxies underestimating quality of life for people with cognitive impairment. How often that bias changes an actual bedside outcome, outside a controlled survey, has not been measured with this precision before, and this paper does not attempt to extend its findings that far.

The paper does not say how often this gap plays out beyond a survey screen. It does say that a signed directive, alone, was not enough to make a dementia diagnosis disappear as a factor in how strangers decided who gets aggressive care. That is a narrower claim than either defenders or critics of advance directives tend to make, and it is the one the data actually support.