clinicians.build · interactive · august 31, 2026

A Thousand People Ask About Golf

Epic's MyChart assistant told a patient it was fine to play golf. The room's first thought was lawyers. Here is the other thought: the same answer, delivered to a thousand adults — and a slider that moves the denominator from the healthiest county in America to the sickest.

Data: CDC PLACES, Local Data for Better Health — County Data, 2024 release
Crude prevalence, adults 18+, modelled from BRFSS 2022 · all 3,144 counties

Eighteen holes is four to five miles of walking, carried out over four hours, mostly by people over fifty. For most adults, that is exercise. For an adult with coronary heart disease, or COPD, or a stroke six weeks ago, it is a clinical question — and the honest answer is usually yes, and here is what to watch for, which is a different sentence from yes.

Below, a thousand dots stand for a thousand American adults. Red dots have the condition. Move the slider and the thousand people are redrawn from a different county.

The condition
median
has the condition inside the 95% interval — could go either way does not
In this county
of adults
Where it ranks
of 3,144 counties
Adults there
18 and over
vs the US
national is the dashed row
Drag the slider. Every position is a real county, ordered from the lowest prevalence in the country to the highest.

All ten conditions at once, for the county you picked

One question, ten different ways it could be the wrong question. Each small grid is a hundred adults in the same county; the number is that county's prevalence, and the pale ring is where the US sits.

These are ten separate estimates. They cannot be added — the same adult appears in several of them.

Why this matters more than the indemnity question

Of 246 leaders polled at Epic's user meeting, 43% expect a patient-facing assistant to be available to most of their patients within the year; 4% say they will not offer one. Every ambient product argued about for two years rested on one sentence: a clinician reads it before it counts. A patient-facing assistant deletes that sentence.

The only thing left between the model and the patient is the escalation threshold — and nobody has published one.

The critical lens: what a pale dot means

The pale dots are not decoration. CDC PLACES county estimates are model-based small-area estimates — multilevel regression on BRFSS survey responses, poststratified onto census demographics. They come with a 95% confidence interval, and for these ten measures that interval averages 4.7 percentage points wide across counties. In the widest county it is 6.4.

The pale band shows ± half that county's average interval. For a county of 5,000 adults, the band can be wider than the difference between that county and the national figure — which is the polite way of saying CDC does not know the county's number to the precision you would need to tune a threshold on it.

The ranking is credible; PLACES is validated for exactly that. The decimal place is not. Anyone who quotes a county prevalence to one decimal without the interval is quoting a model output as a measurement.

What this does not show