Fire on an intensive care unit caused by an oxygen cylinder


  • PH is an employee of BOC Ltd, the manufacturer of the oxygen cylinder described. No external funding or other competing interests declared. Published with the assent of the relative of the patient who was burnt.

This letter is accompanied by an Editorial. See p. 8 of this issue.

We would like to share the information surrounding a recent fire in the intensive care unit (ICU) at the Royal United Hospital (RUH), Bath. As there are ongoing investigations of the event, we must emphasise that the descriptions of the fire and the subsequent events are the recollections of those present at the time. The possible causes of the fire are based on the findings so far of the technical investigation being carried out by BOC Ltd in conjunction with the Health and Safety Executive (HSE), but may not be the actual causes. This letter is not intended as an expert analysis of the event, but is written to inform our colleagues that such an event took place. We also wish to provide some advice to other medical gas users on suggested best practices when using medical oxygen cylinders, in the light of this event and its investigation so far. A final report is awaited from the HSE, which may provide more information.

A self-ventilating patient, receiving facemask oxygen from the pipeline supply, was being prepared for transfer to an ICU at another hospital. A CD-size medical oxygen cylinder was being prepared to administer oxygen to the patient during transfer, and had been placed on the edge of the bed (at the foot end) with the valve pointing towards the head of the bed. The cylinder valve had already been turned on before the incident. A flow of 2 l.min−1 was selected to check that gas was flowing correctly before delivering oxygen to the patient. No tubing had been attached to the outlet at this stage to facilitate this check. Immediately after opening the valve there was an abnormally loud hiss, that appeared to be coming from the top of the cylinder, and sparking was noted coming from the cylinder outlet. Seconds later there followed a very loud bang and four-foot flames came from the valve end of the cylinder. The fire spread rapidly to the bed linen, mattress, curtains and ceiling tiles, and the ICU began to fill rapidly with thick, black, acrid smoke. The burning oxygen cylinder was promptly pushed to the floor by one of the doctors, where it also set the floor covering alight, adding to the smoke. The patient was quickly dragged from the bed by another doctor and a nurse and moved to safety. The fire alarm was activated. Within a few minutes of the fire starting, two other doctors had successfully extinguished the fire, using one carbon dioxide and four water extinguishers. Eleven patients on the ICU were evacuated within approximately seven minutes (with one unstable patient who was receiving controlled ventilation remaining in a side room until an intra-hospital transfer could be arranged). During this time, the entire 15-bedded ICU had become so filled with acrid smoke that visibility was less than a metre, and breathing became difficult. The fire brigade arrived a few minutes later.

The cause of the fire is currently under investigation by the HSE, and it is expected to be several months before the final report is published. From the information received so far, it has not been possible to identify the cause of the ignition as the critical evidence was destroyed in the fire. However, it does appear that the fire was initiated inside the valve and when the regulator (built into the cylinder valve) ignited, it started to vent large volumes of oxygen into the local environment. This had two effects: firstly to cause the bedding to burn violently; and secondly to ignite the plastic guard (fitted to protect the cylinder valve). Whilst the oxygen continued to vent from the cylinder, the fire continued to burn with significant ferocity (Fig. 1). The patient on whose bed the oxygen cylinder lay suffered burns to her lower limbs, and was stabilised and transferred to the tertiary burns centre. All other patients were thoroughly assessed for signs of smoke inhalation, but had suffered no serious harm. All staff were reviewed in the Emergency Department: one member of staff was admitted overnight with smoke inhalation, and one admitted for six hours for nebulisers; all other staff were discharged home that evening.

Figure 1.

(a) The scene following the fire. Note the burnt oxygen cylinder lying on the floor in the foreground. (b) The oxygen cylinder before examination at BOC Ltd. (c) The remains of bedding and mattress after the fire, showing its impact.

Staff from the hospital have met with the author of the interim report (PH) to review its findings and to discuss some of the potential lessons that could be learnt from the incident. It is clear that the cylinder was being used as per the published User Instructions and normal practice. The cylinder had been turned on and the flow selected correctly. However, there were a number of potential actions identified that could have minimised the impact of the ignition had they been taken.

We stress that there are several million medical oxygen cylinders in service in the UK, all filled many times each year, and fires of this nature are very rare. Medical oxygen cylinders are normally very safe to use and handle within the healthcare environment, being designed, manufactured and supplied in accordance with very strict guidelines and standards. However, there is always an extremely small risk when handling and operating high-pressure oxygen cylinders that an ignition might occur. We are aware of two similar ignitions involving medical oxygen cylinders recently in the UK, but as these also are also under investigation any association would be speculative.

The hospital and BOC Ltd have worked together to identify the best practices that should be used when setting up medical gas cylinders and administering medical gases to patients, to minimise the impact of this type of incident (Table 1). These recommendations are currently being integrated into the User Instructions issued by BOC Ltd for use with all cylinders. Cylinder holders to support small medical oxygen cylinders are available but probably not used in most hospitals. National safety organisations, working with medical gas suppliers, may wish to consider these recommendations. Should these recommendations be adopted, changes to both the training and the equipment in many hospitals and other healthcare areas will be required. Specifically, all those involved in handling and using medical cylinders should be trained and be made aware of the potential risks.

Table 1. Suggested best practices when setting up and administering medical gases
1. Set up the cylinder for patient use before placing it close to the patient.

The most likely time for an ignition to occur is either when the valve is initially turned on or when a flow is selected. Hence the advice is to:

a) connect the tubing and oxygen delivery device to the cylinder;

b) slowly open the cylinder valve;

c) select the prescribed flow rate;

d) if required, check the gas is flowing;

f) fit the oxygen delivery device to the patient.

2. Place the cylinder in an appropriately designed holder.

Where possible, cylinders should be placed in holders designed to be fitted, ideally, to the bottom of the bed (or to the back of wheelchairs). The position of the holder needs to take account of how close the cylinder is to the patient. The holder should ideally keep the cylinder upright so that if there is an ignition its impact would be minimised.

Although cylinder holders and brackets are available, a suitable design is dependent on the specific bed or wheelchair being used. As the NHS uses many different types of hospital beds and trolleys, there is no single design that can be used in all situations and this remains an issue to be resolved. BOC Ltd are currently working with bed, trolley and wheelchair manufacturers to develop suitable cylinder supports.

3. Avoid placing the cylinder on the bed next to the patient if at all possible; use extra care when there is no option but to place the cylinder on the bed.There are times when there is no option but to place the cylinder on the bed or stretcher. If this is the only option, setting up and turning on the cylinder before putting the cylinder on to the bed will minimise the potential risk of injury to the patient.

The internal and external investigations are still in progress, and might cast further light on the cause of this event. However, we judge that it is in the interest of both patient safety and the understanding of the wider anaesthetic and medical community that these events are reported as soon as possible. We await the results of the HSE investigation.