Explore
Five views of the same national data. Each answers one question.
Pairs
Pick two markers and see risk of death across both.
Filter by sex and age
Showing all US adults. Pick a group to see its own map; each group’s map is fitted separately.
- eGFR
- 98
- Systolic BP
- 121
- Within data
- yes
Drag the marker, or focus the map and use the arrow keys. Hold Shift for larger steps.
Pick up to three. Each arrow is one study's average change in the marker, from the marker's current position.
Plot your own results over time
Type the date and both values from each of two or more test reports. The map draws your path in date order, and the table compares each step with how much people of your age and sex usually change in a year. Values stay in this browser.
Population associations from US adults, not a personal risk estimate.
What to notice
Risk of death rises steeply as kidney function falls below an eGFR of about 60. Higher blood pressure adds risk everywhere, but most where kidneys work well.
A 30 mmHg rise adds about 40% at eGFR 100 and about 15% at eGFR 60, where risk is already 1.4 times the reference. A report that lists the two values in separate rows cannot show that.
The average adult moves from eGFR 117 and systolic BP 113 in their 20s to eGFR 65 and systolic BP 140 at 80–84. The spread of systolic pressure doubles over that span, from an SD of 11 mmHg to 22.
How to read the map
How to read it
Colour shows risk of death from any cause relative to the reference point. It comes from a Cox model fitted on 46,469 adults with 6,688 deaths (fitted surface). Each contour band is one step on the scale beside the map. Hatched areas hold too few people to estimate.
The read-out gives a 95% confidence interval at the marker: the range the data are consistent with. It reflects sampling uncertainty only, not model or confounding uncertainty. Show uncertainty fades the map, or swaps it for a map of interval width.
Try an intervention draws the average change one study measured in a marker as an arrow from the marker. The study's 95% interval shows as a faint bar or box at the arrowhead. It shows where that change would land on this map, not what the intervention would do for a person.
Up to three arrows can be chained, although effects from separate trials may not add. The interaction test asks whether the two markers combine differently from the sum of their separate effects (likelihood-ratio p = 3×10⁻⁵).
Use Copy link to this view to share the exact position you are looking at.
Biomarker pages: Estimated Glomerular Filtration Rate (eGFR) · Systolic Blood Pressure (SBP).
Ageing
How fast do people drift across the map after 50?
The same people, measured every four years in the US Health and Retirement Study, 2006–2022. The figures are weighted to the US population aged 50 and over.
By age: median and the middle 80%
The population by decade of age (NHANES)
Each curve is the spread of one decade of US adults, 20s at the top, fitted as a normal distribution from NHANES means and standard deviations (log-normal for markers fitted on a log scale). The tick is the mean. 56 markers available. These are population snapshots at one time, not the paths individuals follow.
In the years before death
Mean value by time before death, among people who died during follow-up. Bands are two years wide.
Compare with the Wisconsin Longitudinal Study
Two cohorts, same instruments
The Wisconsin Longitudinal Study measured the same things, with the same instruments, in the homes of 8,925 people in 2011. They were 1957 high-school graduates and their siblings, then aged mostly 64–79. It is one state, mostly white, and unweighted, so it is a check on HRS, not a reference. Medians agree within a few percent for grip strength, waist, BMI and waist-to-height ratio. Peak flow runs about 10% higher in Wisconsin, consistent with a different meter and a healthier-than-average group. Pick Wisconsin in the cohort menu above to see its age curves and its 18-year self-reported weight trajectory.
| Measure | Sex | Age | HRS median (p10–p90) | WLS median (p10–p90) | n HRS / WLS |
|---|---|---|---|---|---|
| Grip strength (kg) | Men | 65–69 | 41.0 (30.5–51.0) | 40.0 (28.2–51.8) | 3,555 / 523 |
| Grip strength (kg) | Men | 70–74 | 38.5 (28.0–48.5) | 38.0 (26.0–48.0) | 3,343 / 2,981 |
| Grip strength (kg) | Men | 75–79 | 35.5 (25.5–45.5) | 33.0 (24.0–42.0) | 2,919 / 247 |
| Grip strength (kg) | Women | 65–69 | 25.0 (18.0–32.0) | 24.0 (16.0–30.0) | 4,812 / 541 |
| Grip strength (kg) | Women | 70–74 | 23.0 (16.5–30.0) | 22.0 (16.0–30.0) | 4,490 / 3,401 |
| Grip strength (kg) | Women | 75–79 | 21.5 (15.0–28.0) | 20.0 (12.0–26.7) | 3,783 / 264 |
| Peak expiratory flow (L/min) | Men | 65–69 | 480 (290–630) | 520 (350–660) | 3,549 / 514 |
| Peak expiratory flow (L/min) | Men | 70–74 | 450 (260–603) | 500 (320–640) | 3,323 / 2,957 |
| Peak expiratory flow (L/min) | Men | 75–79 | 420 (230–570) | 480 (300–633) | 2,933 / 238 |
| Peak expiratory flow (L/min) | Women | 65–69 | 330 (210–420) | 350 (240–440) | 4,912 / 531 |
| Peak expiratory flow (L/min) | Women | 70–74 | 300 (180–400) | 330 (220–420) | 4,590 / 3,345 |
| Peak expiratory flow (L/min) | Women | 75–79 | 275 (160–370) | 300 (200–380) | 3,858 / 256 |
| Waist circumference (cm) | Men | 65–69 | 105 (91–124) | 107 (93–124) | 3,558 / 512 |
| Waist circumference (cm) | Men | 70–74 | 107 (91–124) | 107 (93–124) | 3,319 / 2,971 |
| Waist circumference (cm) | Men | 75–79 | 105 (92–124) | 107 (93–122) | 2,919 / 243 |
| Waist circumference (cm) | Women | 65–69 | 98 (79–119) | 96 (77–121) | 4,826 / 517 |
| Waist circumference (cm) | Women | 70–74 | 99 (81–119) | 98 (80–122) | 4,510 / 3,301 |
| Waist circumference (cm) | Women | 75–79 | 98 (80–119) | 99 (80–121) | 3,818 / 257 |
| Body mass index (kg/m²) | Men | 65–69 | 29.3 (23.9–36.8) | 29.9 (24.3–37.3) | 3,423 / 527 |
| Body mass index (kg/m²) | Men | 70–74 | 29.1 (23.7–36.5) | 29.6 (24.4–36.8) | 3,220 / 3,000 |
| Body mass index (kg/m²) | Men | 75–79 | 28.7 (23.5–36.5) | 29.4 (24.2–35.5) | 2,834 / 248 |
| Body mass index (kg/m²) | Women | 65–69 | 29.3 (22.1–39.1) | 28.5 (22.2–38.3) | 4,702 / 534 |
| Body mass index (kg/m²) | Women | 70–74 | 29.1 (22.2–38.3) | 28.4 (22.7–38.2) | 4,403 / 3,340 |
| Body mass index (kg/m²) | Women | 75–79 | 28.2 (21.5–37.3) | 28.2 (22.5–36.3) | 3,737 / 266 |
| Waist-to-height ratio | Men | 65–69 | 0.61 (0.53–0.72) | 0.61 (0.53–0.71) | 3,526 / 512 |
| Waist-to-height ratio | Men | 70–74 | 0.61 (0.53–0.72) | 0.61 (0.53–0.71) | 3,282 / 2,969 |
| Waist-to-height ratio | Men | 75–79 | 0.61 (0.53–0.72) | 0.60 (0.54–0.71) | 2,882 / 243 |
| Waist-to-height ratio | Women | 65–69 | 0.61 (0.50–0.75) | 0.60 (0.48–0.76) | 4,786 / 517 |
| Waist-to-height ratio | Women | 70–74 | 0.62 (0.51–0.75) | 0.61 (0.50–0.77) | 4,467 / 3,299 |
| Waist-to-height ratio | Women | 75–79 | 0.62 (0.50–0.76) | 0.61 (0.50–0.76) | 3,779 / 257 |
About these numbers
Physical measures from HRS waves 8–16, adults 50 and over, weighted by the physical-measure weights. Age curves are cross-sectional percentiles; change and tracking follow the same people across two visits four years apart.
Only aggregate statistics are published. Any group of fewer than 50 people is left out. How we handle cohort data.
Which test
Which single measurement tells you most, beyond age and sex?
For 56 measurements in NHANES, how much each one improves the prediction of who dies, out of sample, among people of the same age. Grip strength and urine albumin lead; lipids add almost nothing.
Method
Each biomarker enters a Cox model as a restricted cubic spline. Models are fitted on 70% of NHANES 1999–2018 participants and scored on the other 30%. The score is Harrell's concordance: how well the model ranks who died first. It is computed within 10-year age bands, so age alone cannot carry the result.
"Beyond standard risk factors" adds the marker to a refitted panel of systolic BP, BP treatment, total and HDL cholesterol, smoking, diabetes, eGFR and BMI. These are associations in a general adult sample, not a screening recommendation.
Interventions
Which interventions move which markers, and by how much, in trials?
555 measured effects of drugs, diet, exercise and other interventions on specific biomarkers, each linked to its source. They describe study results, not recommendations.
Interventions for stressors themselves (cash transfers, trauma care, caregiver support) are on Models: growth or scarring.
| Biomarker | Type | Intervention | Reported effect | Evidence | Source |
|---|---|---|---|---|---|
| Loading the table… | |||||
Things that worsen a marker
157 diseases, exposures and treatment side effects that moved a biomarker the wrong way in the cited studies.
| Biomarker | Type | Condition or exposure | Reported effect | Evidence | Source |
|---|---|---|---|---|---|
| Loading the table… | |||||
Terrain
Where is the map steep, and where do people settle?
The terrain pages show where risk of death changes fastest for each marker, and the levels the same people return to when measured years apart. They are a separate set of pages.
- OverviewAltitude is fitness, local shape is stability: the idea in one page.Open →
- The mapEvery marker's curve laid side by side as one terrain.Open →
- Every marker's curveWhere risk of death changes fastest for each measurement.Open →
- Where people settleThe levels the same people return to when measured years apart.Open →
- How people moveDrift across the terrain over four years, by age and sex.Open →
- Recovery after a challengeHow quickly a disturbed value comes back, and what that implies.Open →
- Methods and caveatsHow the terrain is computed and what it cannot show.Open →