Steady Hands, Steady Heart. That's the Problem.

745,000 people died in 2016 from working too much.

Not from accidents at work, but from the work itself. Strokes and heart disease, in people whose only real exposure was a long week. The WHO put the number together in 2021, and it was the first time anyone had bothered to count.¹ We know roughly what a marathon does to a heart. We had nothing on what a career does to one.

And the strangest part is that we still don't measure it. There's a whole industry built to tell me what a thirty-minute run did to my resting heart rate, and nothing that tells me what a thirteen-hour shift did. My watch has three hours of my week in high resolution. It has never once seen a shift. The only hours we measure are the ones we chose.

I'm a doctor. Two-thirds of doctors my age are somewhere on the road to burnout. We read the surveys every year, nod, and go back to work.² But a survey is a fact about other people. I didn't want to know what long hours do to doctors. I wanted to know what my sixty were doing to me.

So I decided to measure them.

The tool

Half the tool already existed. The Fitbit gave me a number for every minute of those sixty hours. The climb at 3am could have been a delivery or a nightmare. The flat hour at noon could have been me asleep after a run of nights or a lazy Sunday. Six weeks of data, and no idea what I was doing in any of it.

So I built the other half. Vitals does one thing: it puts a label on the time. Theatre. On-call room. Caesarean at four in the morning. For six weeks I tagged every shift and everything inside it.

Home, a Saturday on-call, and the shifts list: the work logged next to the body's numbers, a line that mostly idles in the seventies, and every day tagged.

It was the first time I had looked.

Hunting for peaks

I'm an obstetrics and gynaecology trainee carrying the emergency bleep, which means the work is whatever comes through the door: a labour that stalls, a haemorrhage, a baby whose heart rate is dropping, a caesarean that has to happen in the next fifteen minutes. Decisions that don't wait, at the hour the body most wants to be asleep. By any ordinary measure, it is a hard job.

So I went looking for the emergencies. I expected to find them as spikes: the crash call, the delivery that turns in seconds, the moment the job showed up in my body.

I couldn't find them. I went back through the six weeks shift by shift, looking for the spike I was sure was in there.

A haemorrhage at three in the morning took me to 94. A crash section, knife to skin inside a minute, 91. A bleed I stood at until it reached two litres, 88.

I climb the stairs to my flat at 96. The hardest my heart worked in six weeks was on my own time, nowhere near a hospital.

And it wasn't only the emergencies. My worst shift of the six weeks was a twelve-hour Saturday on-call with three emergencies in it. My heart went above a hundred for eighteen minutes. Eighteen, out of seven hundred and twenty.

Across six weeks, less than one percent of my working life reached the effort of a brisk walk.

The troughs

And yet I came off those shifts hollowed out. Something was being taken and it wasn't anywhere on the screen.

So I went back through the trace again, this time at the hours between the peaks I never found: ward rounds, notes, the parts of a shift where nothing is happening. That's where I expected my heart to drop back down and stay there.

It never did.

Not once in twelve hours. Not once in six weeks.

A day off next to a normal working day. Green is time at rest.

My heart sat at 78 from the morning handover to the drive home. That is twenty-three beats above my resting fifty-five, small enough to feel like nothing, held for twelve hours. Around sixteen thousand extra beats a shift. Then I slept, got up, and did it again.

I wondered whether it was just being on my feet. But a quiet lunch on a day off settles me into the low sixties. Sitting down to eat on a hard shift, I was still at eighty. Even off my feet, the day wouldn't let me put it down.

Exertion has a shape: up, then down, then recovered. This had no shape. It was a body switched on at eight in the morning and left on.

The two long days. Neither one comes back down.

The nights

Not every shift did it the same way.

Normal days and long days hold the line high from the first patient to the last. Nights look kinder on the graph. There are long green stretches where the heart finally drops. Those stretches are not rest. That's me in the on-call room at four in the morning, spending sleep I don't get to repay.

The sleep I got afterwards didn't repay it. After a night I slept three and a half hours. Less sleep, and less out of it: the deep and dreaming stages, the ones that do the repairing, down by a third.³

So recovery took days rather than hours. After a run of nights my heart rate variability (the gap between beats, which narrows under load) came back within two or three days. My resting rate, the speed it runs at doing nothing, took three or four.^{4 5}

Medicine does not hand out three or four clear days. We come off one block and into the next, and whatever hasn't cleared comes with us.

Feeling fine

I felt fine.

Not every day (nights are nights), but across the shifts the numbers said were costing me most, I felt okay. Often good. I'd come off a run of days sure I was handling it, and my resting rate would still be four beats high three days later.

I can't tell whether I was fine on those days or so used to running low that I'd stopped noticing. Drop recovery far enough, often enough, and low stops registering as anything. It becomes the line we call fine.

That's what stays with me. Not that the job takes something. I knew that. It's that it takes it quietly, somewhere the body stops flagging it, so the one warning I'd trust never arrives.

The wrong hours

I know how thin this is. One person, one rota, six weeks, a strap I bought in a supermarket. It proves nothing. But it's the first honest look I've had at my own job, and I can't unsee it.

Measuring it doesn't fix it. I still work the nights. But a cost nobody can see is a cost nobody has to answer for, and for six weeks I could see mine.

This isn't a medicine problem. It's just the job I happen to have data on. Everyone else is running the same experiment with their own job and no data at all. Decades of it, one week at a time.

We have the instruments. We choose to point them at the safest three hours of our week.

Measure the work.

The tool is Vitals, an app I've been building to log clinical work against wearable data. Every number and chart here came out of it.