One of the pleasures of being a doctor, albeit one who doesn’t see patients, is that you get to chat to other doctors, real ones, about the tricky decisions they have to make. I’ve had two such sets of conversations in the past week that have stuck in my mind.

One was with a junior paediatrician. She talked about being handed a very premature neonate and having to decide instantly whether the baby should be resuscitated. In a split second you make a decision that may well mean that the child lives a life of considerable disability, the parents’ lives are changed forever, siblings are severely affected, and the state is handed a bill of millions. Inevitably the default will be to resuscitate.

I feel deeply uneasy about the seemingly unstoppable trend to keep alive ever more premature babies, particularly in a world where millions of children born at term are dying for lack of clean water and adequate food and of easily treated conditions. But what does it matter what an armchair pundit like me or, come to that, the Archbishop of Canterbury thinks when my young friend is handed a premature baby at 3 am?

Another part of medicine that makes me feel uncomfortable is giving patients with advanced cancer highly expensive drugs that keep them alive for a few more weeks on average. I like to think that I’ll refuse such drugs when my time comes, but I hear that people will do anything to live a few more weeks. Maybe I’ll be the same.

It was thus a great pleasure to find myself at dinner with some blunt speaking and thoughtful oncologists. One of them joked that cancer researchers would soon have to be using stopwatches to measure the extra life given to people by extraordinarily expensive drugs.

“Will dying people do anything to have a few more weeks?” I asked. Interestingly one thought no, the other yes. One quoted research published in the BMJ some 20 years ago that showed that dying people would generally do anything for a few more weeks. But then we debated how the issue might be put to patients.

If you say: “This drug on average will give you two more months” then you might get a different answer from saying: “This drug might give you a 10% chance of living another year or the side effects might mean you die in days.” One oncologist thought that patients tended to hear only the word “year.” The way you frame the information is clearly important.

Much of the research mentions side effects but not resources. If you say: “You could have this drug and it might give you another two months, but it’s so expensive that we might not be able to afford to resuscitate a premature baby” you might get a different answer from when you don’t mention resources. Or in other health systems you might say: “Treatment with this drug is £50 000 a year, and if we don’t give you the drug then your daughter might be able to afford to buy a house” and get another answer altogether.

Doctors generally stay away from talking resources (“money” as they are otherwise known), but we can’t. Doctors should talk prognosis, side effects, and resources—but you can’t do that with an unconscious single mother of a premature baby at 3am.

Richard Smith was the editor of the BMJ until 2004.