The deaths that never get their own line. A data documentary on the mortality of work stress.
A life is about seventy-eight years. Roughly a third of its waking middle is spent at work. This is one life. What follows is a count of the ones cut short by it.
One figure stands for 1,000 people.
The dead being counted here are not soldiers. They are nurses, drivers, founders, engineers. They skew male, middle-aged and older. People who worked.
The numbers already mourned.
US road deaths in 2023: 40,901. Drug overdoses: 105,007. Each is counted every year, named, and tracked by a federal agency. Hold them as a yardstick.
Work stress: more than 120,000 a year.
More than the roads. More than US drug-overdose deaths. About a full airliner of people. Every single day.
745,000 worldwide, in a single year.
From long working hours alone. This counts one cause and two diseases, stroke and heart disease. Everything else is on top of it. This is the floor.
Through the body, over years.
Stroke: about 398,000. Heart disease: about 347,000. The dose follows the hours: a 55-plus hour week carries about a third more stroke risk than a normal one.
A death is the part that gets counted.
It is also the smaller part. The same long hours that killed 745,000 in 2016 also erased 23.3 million healthy years of life: the strokes and heart attacks survived but never recovered from. Most cardiovascular events are survived. A stroke is far likelier to disable than to kill. The toll that can be counted is the smallest version of the toll there is.
It is not effort. It is control.
Sort the same civil servants by how demanding their jobs were, and the risk barely moves. Sort them by how much control they had, and it nearly doubles: an odds ratio of 1.93 for the least control. The executives had the most demanding jobs and the most say, and the lowest risk. The danger is effort you cannot steer.
At this scale, it stands with the largest causes of death.
Set more than 120,000 beside the CDC's 2023 causes of death and it belongs among the leading ones, by size. But size is not the same as a place on the list.
It isn't long hours. It's the whole shape of a bad job.
Losing the job, or holding one with no safety net, raises the risk of dying the most. Low control, insecurity, shift work, long hours sit in a modest middle. The toll is not one heroic number. It is how many people stand under the modest ones at once.
Stress is not a metaphor in the arteries.
If stressed people simply had worse habits, behavior would explain the harm. It does not. Trace the effect to its mechanisms, and only about a third runs through sleep, smoking, and inactivity. The other two-thirds is the body itself: blood pressure, cortisol, the metabolic syndrome that runs ahead of heart disease.
Score it in the open, including the box it fails.
Against Bradford Hill's checklist, the one that pinned smoking to lung cancer, work stress is strong where it is hardest to fake: the stress comes years before the heart attack, across dozens of cohorts, though the clearest are European and the US results are weaker. It is weak on strength, the risks are modest. And one box stays empty: no trial has ever lowered job strain and counted the heart attacks. The honest verdict: work stress is "associated with a moderately elevated risk," not a proven cause.
The crowd still standing in it.
Worldwide, 41% of workers felt a lot of stress yesterday. Only 23% are engaged in their work. One in five felt lonely. The exposure in every chart you just saw is the daily condition of most of the people you work with.
The lever is real. Its size is not yet counted.
The two natural experiments point both ways. When people lose their jobs involuntarily, heart attacks and strokes roughly double. When one workplace was deliberately improved in a randomized trial, the benefit showed only among those already at risk, with no clear effect overall. The lever looks real. How many lives it moves has not been counted.
Control. Limits. Support.
The harm was never the effort itself. It was its shape: demand you cannot steer, hours with no end, support too thin. The evidence keeps naming the same three as protective, each with a name, an effect size, and an owner. The link is association, not a coroner's verdict. But it is large, it climbs with the dose, and it has held across decades and continents.
The full No. 3, with every chart and every citation.