Counted Health · Invisible Illness Atlas ← back to the Atlas
How we know it’s real

The evidence, and how we ruled out “it’s just age”

A map is only worth trusting if the pattern holds up under hard questions. Here is what we measured, how we built the burden groups, and the checks that show the gradient is real — not just older people, and not a fluke of the data.

What we measured

Everything is built from a large, trusted U.S. government health survey called NHANES (the National Health and Nutrition Examination Survey). For each person, NHANES collects far more than one doctor usually sees at once. We use three kinds of information from the same person:

Normally these live in separate places, so no one sees them together. We combine them into a single burden score for each person — the higher the score, the more these signals are off at once.

How we sorted people into five groups

We line everyone up by their burden score, lowest to highest, and cut the line into five equal groups (statisticians call these “quintiles”). Group 1 is the lowest burden; group 5 is the highest. These are simply groups along a scale — not diagnoses, and not clusters the computer “found.” We chose the sorting up front so it can’t be tuned to produce a nice-looking answer.

The key test: is it just age?

The obvious objection to any “sicker group” finding is that the sicker group is just older, and older people have more chronic illness. So we tested exactly that.

33% → 72%
Share with a diagnosed chronic condition, lowest group to highest (adjusted to reflect the whole country).
2.5×
The highest group is about 2.5× as likely to have a chronic illness as the lowest — after accounting for age, sex, race and weight.
Holds up
95% confidence range 1.9–3.1; the odds this is chance are less than 1 in a billion.

In plain terms: even when we compare people of the same age, sex, race, and body weight, the highest-burden group still carries far more chronic illness. Age matters too — it is a real co-driver — so we say the burden gradient is independently associated with chronic illness, not that it is independent of age. Being honest about that is the point.

The pattern is even cleaner for women: from 36% in the lowest group to 77% in the highest — the group these illnesses tend to hit hardest.

It shows up across age bands, too. Younger adults have lower rates overall, but the same climb from low to high burden. In the oldest group the rate is high everywhere, so the gradient flattens — a ceiling effect, and exactly what you’d expect if the score is measuring something real.

Age groupLowest burdenHighest burden
20–3922%48%
40–5932%69%
60+73%84%

“People like you”

A single national average can hide a lot. A level of fatigue that looks “normal” against everyone may be clearly high for a young woman. That is why the tools let you compare against your own age and sex, not just the average American — so the burden a general “normal” would hide becomes visible.

What this is — and isn’t

Technical detail
Numbers come from a design-aware, survey-weighted logistic regression on the NHANES 2013–2014 tri-modal complete-case adults (n = 3,919), with cluster-robust standard errors by survey design unit. Outcome: any self-reported doctor-diagnosed chronic condition. Highest-vs-lowest quintile adjusted OR = 2.45 (95% CI 1.92–3.12, p = 4.4e-13), adjusting for age, sex, race/ethnicity, and BMI. Per-quintile trend OR = 1.23; per-SD burden OR = 1.41; age alone OR = 2.19 per SD (a genuine co-driver). The full script, data manifest, and validation outputs are kept in a self-contained, reproducible source project (available to reviewers on request).