Dr Daniela Gentile, BSc DipMedTox PhD

Defining the true epidemiology of CO exposures is challenging due to variations in how data is collected, categorised and recorded.

Common sources of CO include house fires, defective heating appliances or generators, vehicle exhaust emissions and from cigarette smoke. Exposures may be acute or chronic and can occur either unintentionally or intentionally through an act of self-harm. Unintentional exposures may be further subdivided into those related to fires (where additional toxicity such as cyanide may contribute) and non-fire related CO exposures.

Unintentional non-fire related CO exposures pose a serious public health challenge and as such have been the primary focus of our study. Patients are often unaware of the presence of the poisonous gas even after they begin to experience symptoms. Public health policy is focused on raising awareness of this hidden danger while identifying and eliminating potential sources of CO.

The diagnosis of CO poisoning depends on confirming a history of exposure, identifying symptoms consistent with CO poisoning, and demonstrating an elevated carboxyhaemoglobin concentration. A variety of methods can be used to measure this however blood sampling with spectrophotometrical measurement (defined as a percentage of total haemoglobin (COHb%)), remains the gold-standard method.

Taking blood
Taking blood - blood sampling with spectrophotometrical measurement remains the gold-standard method

The diagnosis, however, is not always straightforward as symptoms are varied and non-specific, may be affected by age and co-morbidity, and often correlate poorly with COHb%. Additionally, COHb% may be difficult to interpret as it is dependent on the atmospheric concentration of CO at the scene of exposure, activity levels of the patient following exposure, whether the patient is a smoker or has been in an enclosed environment with smokers.

Furthermore, the half-life of COHb% is reduced from approximately 5-hours when breathing air to 80-minutes following administration of 100% oxygen, therefore the timing and administration of any supplemental oxygen will further complicate interpretation. Finally, it has been suggested that as individual doctors encounter CO poisoning so infrequently, CO exposure may not be immediately considered at the time of presentation with any further delays having implications for further variations in COHb% measurements.

The UK National Poisons Information Service (NPIS) provides information and evidence-based management advice to healthcare professionals through the online poisons database, TOXBASE® and a national 24-h telephone line, staffed by poisons information specialists and supported by consultant clinical toxicologists.


This service is used by front line health professionals when managing poisoned patients, including those exposed to CO. It is therefore uniquely positioned to gather data from healthcare professionals across the UK to help to understand our experience of CO poisoning in the UK. Through a greater understanding of the scale of the problem, symptoms and sources of CO, recognition and management of these patients may be improved to allow targeted treatment towards those most in need while preventing unnecessary hospital admissions.

The Research Project

The research project was led by the NPIS Edinburgh unit with the following aims:

  • To assess methods to obtain exposure details from clinicians contacting the NPIS by telephone or using the TOXBASE database.
  • To gain a greater understanding of the epidemiology of carbon monoxide poisoning to improve its prevention, diagnosis and management in the UK.
  • To obtain more specific details of the demographics, symptoms reported, sources of CO and results of investigations.
  • To improve recognition and management of these patients to allow targeted treatment towards those most in need while preventing unnecessary hospital admissions.
  • To improve the recording of carbon monoxide poisoning in the UK across the different healthcare professions, facilitate a more comprehensive and robust analysis of the true number of non-fatal poisonings in the UK, and assist in the reduction of misdiagnoses of CO poisoning.

The Study

We analysed all CO-related enquiries to the NPIS (telephone calls and TOXBASE accesses) between 01 July 2015 and 30 June 2019. Telephone enquiries to the NPIS are routinely recorded in the UK Poisons Information Database (UKPID).

Enquiries were received from hospitals, primary care, NHS triage services (e.g. NHS 111/Direct/24) and the ambulance service. The UKPID database was interrogated to ensure all potential CO exposures were identified. Data collected included patient demographics, location and source of exposure, symptoms reported, exposure severity and COHb% concentration where available.

Medical professional
Medical professional - enquiries were received by healthcare professionals

In addition, accesses to the TOXBASE CO management page were interrogated during the same period. An online questionnaire was attached to TOXBASE which appeared on the screen at the point of access, inviting users to anonymously provide relevant details about the exposure as they were treating the patient.

Follow up questionnaires were sent to all enquirers in an effort to capture as much data as possible as some information may not have been available at the time of the initial enquiry. Enquiries where the CO exposure was fire related (e.g. house fire) were excluded from the data analysis and intentional CO-exposures (self-harm) were analysed separately.

The Results

This study presents data on a total of 4323 unintentional non-fire related CO exposure enquiries to the NPIS related to between July 2015 and June 2021.

These were most often reported during winter months, involved all age groups but most frequently children and adults aged 20–39 years. They commonly occurred in the home as a result of faulty boilers.

Reassuringly, the majority of cases were of low severity, however eleven fatalities were reported. Symptoms were typically non-specific with the CNS being most frequently implicated e.g. headache, fatigue.

Where COHb% concentrations were available, the data suggest a positive correlation with poisoning severity, however this was not statistically significant.

Where the information was volunteered, activation of a CO alarm was reported in 20.8% of cases. The majority of these cases were associated with mild symptoms or no symptoms at all.

There were 225 exposures reported as intentional (self-harm). The majority involved vehicle exhausts. Relative to unintentional exposures, a greater proportion of self-harm exposures caused severe toxicity. There were nine fatalities.

There is a real challenge with variation in how data regarding CO exposures is recorded, and reported. Hospital Episode Statistics (HES) data (2015-2018) reported 208 admissions/year due to accidental CO exposures in the UK, while literature reports an estimated 4000 ED presentations annually. We have presented data on 4323 NPIS enquiries (approximately 720/year) which represents information from healthcare professionals treating CO poisoned patients in all four UK nations.

Importantly, the data presented here includes information on clinical parameters including biomarkers, poisoning severity and the source and location of the exposure, not traditionally included in other epidemiological studies using coded hospital data. Furthermore, in addition to admitted patients, the current study includes patients presenting to EDs who are discharged without admission and those assessed in the community who do not always attend a hospital.

There are a number of limitations associated with this study. Similar to other poison centre studies, these data represent only cases where health professionals have contacted the NPIS for clinical management advice and therefore may underestimate the true incidence of CO poisoning in the UK. Minor exposures where the patient does not present for assessment or a clinician does not require NPIS advice are not included. Similarly, severe exposures resulting in out of hospital deaths would not be included.

Future Research

Following acute (short term) exposures, as patients are usually discharged without being admitted to hospital, there is little information regarding longer term complications (if any). Similarly, there is limited information on chronic, low level CO poisoning, therefore it would be interesting to investigate if any ongoing sequelae are reported following these exposures.

Dr Daniela Gentile has been leading this project at the NPIS. To hear more about this study sign up to the presentation on Thursday 20th January at 3pm.