Capnography carries risk information the blood pressure does not

Design

Retrospective single-centre cohort study with multivariable logistic regression

Population

185,455 adults undergoing non-cardiac surgery under general anaesthesia with mechanical ventilation, 2010 to 2020

Primary outcome

In-hospital mortality

Effect

Adjusted OR 1.63 (95% CI 1.36 to 1.86) per 5 mmHg decrease below median end-tidal CO2; overall mortality 0.85%

One hundred and eighty-five thousand adults having non-cardiac surgery under general anaesthesia with mechanical ventilation at a single tertiary centre over a decade were analysed for the relationship between mean intraoperative end-tidal CO2 and in-hospital death. Lower values were associated with higher mortality, non-linearly, at an adjusted odds ratio of 1.63 for every 5 mmHg below the median.

The design point is the adjustment. Low end-tidal CO2 is usually explained away as either over-ventilation or poor perfusion, so the analysis adjusted for minute ventilation and for the severity and duration of intraoperative hypotension. The association survived both, and there was no interaction with hypotension — the signal is not simply a re-reading of the arterial line.

What a low value is marking is most likely pulmonary blood flow, which is to say cardiac output, which no routine monitor in most theatres measures. That makes the capnograph trace a perfusion monitor that is already on every patient. It is an association from one centre's records and should not be turned into a target; a number to act on is not the same as a number to chase.

• Treat a falling end-tidal CO2 with unchanged ventilator settings as a perfusion question, not a ventilation one.
• Check the trend against minute ventilation before adjusting either.
• Do not ventilate to a CO2 number; nothing here shows that raising it helps.
• Note the value in the postoperative handover for a patient with a persistently low trace.
• Remember the sensitivity of the trace to cardiac output is what makes it useful in low-resource theatres without advanced monitoring.

Why it matters

The monitor every anaesthetised patient already has may be reporting cardiac output that nothing else in the room is measuring.

Don't overread it

This is a retrospective association at one centre; treating end-tidal CO2 as a target has never been tested.

The statistics, in plain English

An adjusted odds ratio of 1.63 per 5 mmHg sounds dramatic against a baseline mortality of 0.85 per cent — in absolute terms the shift is small for any one patient, and the value of the finding is in risk stratification rather than in individual prediction. Adjustment can only account for confounders that were measured: sicker patients have lower cardiac output and also die more often for reasons the model never sees. The non-significant interaction with hypotension (P=0.19) means the two act separately in this dataset, not that they are biologically independent.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.