The deprivation gradient

Where the gradient bites — and where it doesn't

Complete tooth loss is three times higher in the most-deprived tenth of neighborhoods. Binge drinking is the one measure that runs the other way.

Sort America's neighborhoods into ten bins by the Area Deprivation Index and walk from the least-deprived tenth to the most-deprived. How much worse does each measure get? Putting all 26 on a single relative scale — each line starts at 1.0× in the least-deprived decile — makes the answer legible at a glance: 6 of the 26 measures at least double by the time you reach the most-deprived tenth.

The deprivation gradient, all 26 measures at once

Burden relative to the least-deprived decile (log scale) — hover a line or label to isolate it

Population-weighted means per ADI national-rank decile. The steepest line is complete tooth loss among seniors: from 8.55% in the least-deprived decile to 26.31% in the most-deprived — 3.08× higher. Utility-shutoff threat (2.8×) and smoking (2.3×) follow.

Teeth are the body's deprivation index

It is fitting that the steepest gradient belongs to dentistry. Complete tooth loss is cumulative, cheap to prevent, expensive to treat, poorly covered by insurance, and almost perfectly classed — which makes a senior's smile one of the most legible records of a lifetime of neighborhood deprivation that epidemiology has.

What the gradient leaves alone

Three lines barely move, and each is a lesson. Cancer prevalence (0.96×) is nearly flat because it mostly tracks age, and deprived neighborhoods skew younger — flatness here is an artifact of who lives where, not equity in cancer risk. Loneliness (1.06×) is almost evenly distributed: the most isolated places in America are not the poorest, they are the youngest and densest. And binge drinking runs backwards (0.81× — that is, 19% lower in the most-deprived decile), the one health behavior that rises with affluence.

The gradient, in other words, is not a law of nature that drags every measure equally. It bites hardest where prevention is priced out of reach — and it spares the burdens that answer to age and density instead of income.

Read this carefully. Estimates are CDC PLACES-style model-based small-area estimates, not direct measurements. Every association here is ecological — it describes places, not people, and implies nothing about causation. Cross-measure models are fit on the ~23,800 ZIP/ZCTA areas with complete data on all 26 measures (coverage is limited mainly by the newer social-needs measures); maps and community-type assignments extend to areas observing at least 18 of the 26. Full details on the methods page.