Recognising Carbon Monoxide Poisoning: The Tells That Actually Work
The most useful sign is nearly specific: symptoms clear when you leave the building and return when you come back — no viral illness does that. Also multiple people and pets affected at once, symptoms tracking a combustion appliance especially at the first cold snap, and the order of collapse (pets, then children, then adults). Recognition can only ever be backup, because CO degrades the judgement needed to notice it and victims routinely misattribute symptoms.
You cannot sense carbon monoxide, and it degrades the judgement you would need to identify it. Recognition is therefore a **backup** to a working detector, never a substitute. With that stated, several patterns are genuinely diagnostic. ## The tells that work **Symptoms clear when you leave and return when you come back.** This is the single most useful sign, and it is nearly specific. No viral illness behaves this way. Anyone who feels ill indoors and noticeably better after some minutes outside — and then declines again on returning — should treat CO as the leading hypothesis. In documented survival accounts this is the observation that broke the case, sometimes performed accidentally. **Multiple people and pets affected simultaneously.** A whole household with a "flu" that arrived the same evening is a pattern worth distrusting, especially with animals affected too. **It tracks a combustion appliance.** Symptoms that begin with the first cold snap when the heating starts, or that coincide with an appliance running, point at the source. A cracked heat exchanger dumping combustion gas into the supply air is a classic mechanism and is invisible without inspection. **The order of collapse.** Pets go down first, then children, then adults. Foetal haemoglobin binds CO even more avidly than adult haemoglobin. A pet that is unusually lethargic in a house where someone has a headache is a signal, not a coincidence — the same principle as the coal-mine canary, which was a faster casualty rather than a sensor. See Why No Animal Can Sense Carbon Monoxide. ## Why recognition cannot be the primary defence CO causes drowsiness, confusion and impaired reasoning **early**, and it strips out exactly the cognitive resource required to notice it. The failure mode is not ignorance but misattribution: victims routinely construct elaborate innocent explanations for their symptoms — flu, hangover, exhaustion, stress — and there are documented cases of poisoned people reaching genuinely bizarre conclusions rather than suspecting the gas. Two things improve the odds within those limits. **A prior hypothesis.** Someone who has been told about CO repeatedly has the possibility available for recall under impairment, which is when generating a novel hypothesis is hardest. **Attribution clarity.** A symptom with no innocent explanation available demands a cause. This cuts both ways in an underappreciated manner: a person who habitually feels hungover or foggy for ordinary reasons has an innocent explanation ready to hand, and will use it. Which is why acclimatisation is so dangerous — it removes the anomaly. See Carbon Monoxide Poisoning: Why You Cannot Build Useful Tolerance. ## What to do **Get out. Stay out. Call emergency services.** Fresh air stops further uptake, and carboxyhaemoglobin still takes 4–5 hours to clear on room air, so "air out the house and go back to bed" is how people relapse. High-flow oxygen cuts that to under an hour and is the reason to be assessed rather than to self-manage. Significant exposures carry a 10–30% risk of delayed neurological sequelae appearing days to weeks later, so medical assessment is warranted even when everyone feels recovered. See Carboxyhaemoglobin: Binding, Half-Life, and Reference Levels. ## Prevention, in priority order 1. **Sealed ten-year-battery detectors on every level.** This eliminates the commonest failure — a dead battery someone meant to replace. Sensors expire in about 5–7 years regardless. 2. **A low-level monitor** if anyone in the household is infant, elderly, pregnant or cardiac, or if there are unexplained chronic symptoms — standard alarms are silent by design below roughly 30 ppm. See Electrochemical CO Detectors and the UL 2034 Time-Delay Curve. 3. **Annual heating inspection**, specifically the heat exchanger for cracks and the flue for blockage. Yellow rather than blue flames, soot deposits, and unexplained window condensation are warning signs. 4. **Never run an engine in an enclosed or compromised space.** Vehicle cases include people sheltering in a stationary car — a blocked or damaged exhaust can fill a cabin with lethal concentrations while the occupants simply feel sleepy.