Most patients who
deteriorate in hospital
are visible first.

She was struck on the head by a golf ball. She was taken to Royal North Shore Hospital in Sydney, where she was treated for a fracture of the skull — an injury that was not, in itself, expected to kill her.

She died in hospital, of respiratory arrest, caused by the depressant effect of the pain medication she had been given.

Nothing exotic killed her. What the coronial inquest found instead was a sequence of ordinary things that did not connect: an observation that was not escalated, a medicine that should not have been given, a handover that did not carry the risk, a record that never held the whole picture. Every piece of what was needed to save her existed somewhere in that hospital. None of it arrived in the same place at the same time.

“…not enough doctors, not enough nurses, inexperienced staff, poor communication, poor record keeping and poor management. These are systemic problems that have existed for a number of years and regrettably they all surface in the death of Vanessa Anderson.” NSW Deputy State Coroner · inquest findings · 24 January 2008

Her case helped precipitate a Special Commission of Inquiry into acute care in NSW public hospitals, and through it, the state-wide escalation system that every NSW hospital runs today. Two decades on, her inquest is still the clearest short description of how a modern hospital fails a patient it could have saved.

Dylan Thomas sent nineteen lines to an editor in Rome. He had written them for his father — a Swansea schoolmaster, ill for twenty years, losing his sight, and dying. They became some of the most quoted lines in the language: Do not go gentle into that good night. Rage, rage against the dying of the light.

In the postscript of the letter that carried the poem, he added one sentence.

“…the only person I can't show the little enclosed poem to is, of course, my father, who doesn't know he's dying.” Dylan Thomas to Marguerite Caetani · 28 March 1951

The man the poem was written for had never been told. He died the following December, his son holding his hand. The rage in those lines is what was left over when the conversation did not happen.

This is the second failure this platform is built around, and it is not Vanessa Anderson's. Hers was a death that should have been prevented. This is a death that should have been spoken about — a decline visible for years, in front of everyone who loved him, and silence where the conversation belonged.

A hospital gets both kinds wrong for the same underlying reason: something was knowable, and nobody acted on it in time. That a patient is dying is a prediction like any other, and recognising it early buys the one thing that cannot be bought late — time to talk, while there is still someone to talk to.

It is not a rare problem. About a third of rapid response calls reach a patient who is dying rather than one who can be rescued — a finding from our own trial data. An emergency service is arriving, at speed, where a conversation was what was needed.

And that is only the half that gets noticed. Most people approaching the end of life never have a sudden deterioration at all. They decline slowly, across weeks and admissions, and never cross the threshold that summons anyone — so a system built to catch rapid deterioration never sees them, and the conversation comes late, or not at all.

So this platform is built to hold two questions at once, and to tell them apart. Is this a deterioration that can be reversed? Or is this someone near the end of a long decline, where the right answer is not a rescue but a conversation? At the bedside, at three in the morning, those can look alike — and they call for opposite things. Separating them is among the harder problems in this field, and meeting it is a deliberate part of the design, not a by-product of predicting deterioration well.