A study of 7,415 people in the NIH's All of Us Research Program suggests that stress translates into measurable biological wear, allostatic load, more consistently in people whose household income sits above the study's median than in those below it, and the researchers' explanation for the gap has nothing to do with resilience

At a health psychology conference in Pafos earlier this month, I sat through a roundtable that spent an hour on a question I didn’t expect to still be open at this stage of my field: how do we know the things we measure actually exist?

Not in a mystical sense. In the ordinary sense that a lot of psychology, mine included, runs on composite scores built out of smaller measurements that each individually mean very little, stacked together into a number we then treat as if it were as real as a temperature reading.

I do work adjacent to exactly that problem, building objective, measurable markers of emotional and physiological states out of things like behavior and expression rather than self-report.

So when I came across a new study built on one of the older, more established composite scores in health research, allostatic load, and found that it behaves differently depending on how much money is in a household, I recognized the shape of the argument before I finished the abstract. The real story turns out to be about what a measurement can and can’t see once the thing it’s measuring is already maxed out.

A word for the tab your body keeps

Allostatic load is the term researchers use for the cumulative physiological cost of managing stress over time, the wear that accumulates across the body’s regulatory systems when they’re asked to adapt, again and again, to demands that never fully let up.

It isn’t measured with one test. A recent PLOS ONE study, led by researchers at Weill Cornell Medicine using data from the National Institutes of Health’s All of Us Research Program, built its version of the score from twelve biomarkers spanning cardiovascular function, cholesterol and triglycerides, blood sugar, kidney function, immune markers, and waist-to-hip ratio.

No single biomarker is dramatic on its own. A single elevated number, taken alone, tells you almost nothing. It’s the pattern across all twelve that’s supposed to reveal something a single lab result can’t: not that a person is sick, but that their body has been working overtime for long enough to leave a trace.

The pattern that should have shown up everywhere, and didn’t

The study is a single, cross-sectional analysis, a snapshot of stress and biomarkers measured at one point in time rather than tracked over years, so it can show association, not cause. With that caveat in place: it drew on 7,415 adults enrolled in All of Us and looked at whether self-reported stress predicted allostatic load the same way across income levels. Across the full sample, higher perceived stress was reliably associated with higher odds of elevated allostatic load. But when the researchers split participants by household income relative to the federal poverty line, using the sample’s own median as the dividing point rather than the poverty threshold itself, stress stopped behaving consistently.

Above the median, stress and biological wear moved together about as strongly as you’d expect. Below it, that relationship largely flattened out. I want to flag honestly that the authors describe this split as suggestive rather than statistically decisive, with a heterogeneity value that falls just short of the conventional threshold researchers use to call a difference confirmed. It’s a real pattern worth taking seriously. It is not, on its own, proof.

What makes the finding matter is the direction it points. If stress were simply less damaging to people with less money, that would be one story, and an uncomfortable one to tell responsibly. That is not what the researchers argue happened.

The researchers had a simpler explanation sitting right there, that poorer bodies are just more resilient to stress. They chose not to use it.

Not resilience, something closer to already full

The explanation the researchers offer is a ceiling effect: a system already so taxed that one more stressor barely registers. 

In the paper’s own words, people below the median are “more likely to be exposed to multiple adverse health determinants and increased biological risk factors, and thus the singular impact of one of these determinants on AL is likely to be less pronounced.” One additional stressor, in other words, barely moves a needle that a dozen other unmeasured stressors have already pushed most of the way. The paper is explicit that this reflects “inherent inequity between groups, rather than diminished risk” for people with less financial cushion. Their bodies are not handling stress better. Their allostatic load is already elevated by so many overlapping pressures, financial precarity, unstable housing, discrimination, unsafe or under-resourced neighborhoods, that any single additional stressor barely registers as a new signal in a system already running near its ceiling.

That distinction matters more than it might look like on the page. A resilience story lets everyone else off the hook: if lower-income bodies are simply tougher, there’s nothing structural left to explain. A ceiling story does the opposite. It says the measurement itself is behaving exactly as you’d expect a maxed-out instrument to behave, and that the flat line isn’t good news wearing a strange shape.

What I keep coming back to from that roundtable

The reason the Pafos conversation stayed with me is that allostatic load is precisely the composite construct that discussion was warning against treating as unquestionably solid. Twelve biomarkers, one composite number, and a lot riding on the assumption that the number means the same thing in every body it’s calculated from. This study is one of the more careful examples I’ve seen of researchers taking that assumption seriously rather than past it: instead of concluding that stress just matters less for people with lower incomes, they asked whether their instrument might simply be less sensitive in a population where the baseline is already elevated. That’s a more honest, and considerably less comfortable, place to land.

I’ve struggled with anxiety for most of my adult life, and one of the things that struggle taught me is how easy it is to mistake a body that has stopped signaling clearly for a body that’s coping fine. A flat reading isn’t always relief. Sometimes it’s just a system with nothing left to register the next thing being asked of it.

If any part of this is landing closer to your own life than it is interesting as a study, that’s worth paying attention to on its own terms, and talking to someone, a doctor, a therapist, someone you trust, tends to help more than any single number a study can put on what your body has been carrying.

Note: This piece is about what the research design does and doesn’t support, not a diagnosis of what any individual set of lab values means for any individual body.