General anaesthesia

How can a surgeon be confident a patient is unconscious when nobody can measure consciousness directly?

Millions of people a year are rendered insensible so that someone can cut them open. The anaesthetist writes a number in the chart and the surgeon proceeds. But consciousness is the one thing nobody can inspect from outside: there is no probe that reads experience. So on what does that confidence rest — and what happens when the substitute and the thing come apart?

Start by noticing that "asleep" is the wrong frame, and that it hides the problem. Sleep is one state, and one you can be shaken out of. What the surgeon needs is at least four separable things: that the patient has no experience, that they form no memory, that they do not move, and that heart and circulation stay stable under the knife. It would be convenient if one dial governed all four. It does not.

The clearest evidence comes from where the drugs act. Immobility during surgery turns out to be largely a spinal effect, not a cerebral one. In goat experiments where the anaesthetic reached the brain preferentially while the body stayed comparatively unexposed, the concentration needed to abolish movement rose sharply — the cord, not the cortex, was holding the animal still. Amnesia, meanwhile, comes cheap: memory formation is knocked out well below the concentrations that abolish awareness.

Now look at what is actually measured. The standard yardstick for an inhaled agent is MAC — the minimum alveolar concentration at which half of patients do not move in response to a standard surgical incision. Read that again. It is calibrated on movement, in a population, at the fifty percent point. It says nothing directly about anyone's inner state, and it is not even a threshold for the individual in front of you.

So why does it work? Because of a fortunate margin. The concentration at which half of people stop responding to a spoken command — MAC-awake — sits at roughly a third of MAC. The dose that stops movement is comfortably above the dose that stops experience. A patient who is still is therefore very probably unaware, not because stillness means unawareness, but because the two thresholds are ordered, and one is far easier to see.

Which exposes the weak point at once. Give a muscle relaxant and the movement signal is destroyed: the most informative proxy is switched off precisely so the surgeon can work. Reported awareness clusters accordingly — in paralysed patients, in cardiac and obstetric cases, and at induction and emergence, when drug levels move fastest.

How common is it? The numbers are instructive because they disagree with each other. A UK national audit found spontaneous patient reports at roughly one in nineteen thousand general anaesthetics. Studies that instead ask every patient in a structured interview find something like one or two per thousand. And the isolated forearm technique — a tourniquet applied before the relaxant, leaving one hand free — has found around one in twenty patients squeezing on command shortly after intubation, almost none of whom remember it. Three methods, three answers spanning three orders of magnitude, because each asks a different question: what is complained of, what is recalled when prompted, and what responds at the time.

The obvious fix is to measure the brain, and monitors exist that reduce the EEG to a single index with a target band. But that index was built by fitting an algorithm to a database of patients given particular drugs. Ketamine and nitrous oxide, which deepen anaesthesia, push it up; the monitor tracks a drug's EEG signature rather than consciousness as such. Tested head to head, an index-guided protocol did not beat the cruder tactic of alarming when the end-tidal agent concentration drifted low.

The tempting misreading is that depth of anaesthesia is a single quantity we are merely bad at measuring. It is better understood as several effects produced together by the same drugs, at different sites, in a reliable order. Confidence comes from that ordering, not from observing consciousness — which is why the practice survives the problem of other minds rather than solving it.

Two honest qualifications. The mechanism of anaesthetic unconsciousness is genuinely unsettled: whether it is best described as a breakdown of cortical integration, a thalamic gate closing, or something else is still argued. And "awareness" is not one phenomenon either — responding to a command during surgery, recalling something afterwards, and being distressed by it years later are three outcomes with three frequencies, and conflating them is how the incidence figures come to differ so wildly.

This also differs from the ordinary problem of a test standing in for what it cannot see. The difficulty is not a proxy imperfectly correlated with a construct that could in principle be checked; it is that the target is unobservable from outside in principle, so the proxies can never be validated against it — only against each other.

• Why depth indices behave so differently for ketamine and nitrous oxide than for the agents they were fitted to.
• What distinguishes patients distressed years later from those who recall an episode with indifference.

Every term the collection defines is gathered in the glossary.