Subjecting infants to low oxygen concentrations seems unethical

EDITOR—The ethics of Parkins et al's study depend partly on whether there is any potential benefit to the infants from the experiment.1 If parents subject their infant to low oxygen concentrations in an aeroplane they take a risk which is balanced against the benefit of air travel. Unless there is some real prospect of identifying infants at risk and offering useful protection against the sudden infant death syndrome, it seems to me to be unethical to subject those infants to the risk of exposure to low oxygen concentration. The fact that some of the families had previously experinced the loss of a child may have increased the likelihood of compliance. This makes me even more worried about the ethics of the study.

Research should contain element of treatment

EDITOR—Parkins et al and Hughes, the chairman of the local ethics research committee, in their reply to the commentary by Savulescu, repeat that the parents knew of the potential risks and gave consent.1 This does not justify the research. In the case of proxy consent the people—in this case the parents—vested with that power must use it reasonably. The law regarding the participation of children in non-therapeutic research is unclear. The most favourable suggestion, from the researchers' point of view, is that the validity of consent should be based on whether the parents are clearly not acting against the best interests of the child. Thus the amount of risk that the child is exposed to becomes relevant.2

According to the British Paediatric Association it would be unethical to submit child subjects to more than minimal risk when the research offers no or minimal benefit to them.3 The Institute of Medical Ethics has defined minimal as a risk of death of less than 1:1 000 000, a risk of major complications of less than 1:100 000, and a risk of minor complications of less than 1:1000.4 Parkins et al's study would have to be repeated almost 3000 times without major complication for the risk to be classified as minimal. The authors report an oxygen saturation of <60% for 30 seconds in one child and of <80% for over two minutes in another. Incidents such as these would be classified as major complications by anyone.

Much of the criticism could have been shortened if the study had been undertaken in infants who were to travel on aircraft, as suggested by Savulescu. This would have introduced an element of treatment. Collaboration with an airline would have made the study more cumbersome to undertake, but the convenience of the researchers cannot be used as an excuse for the participation of vulnerable volunteers.

Perhaps I am horrified by this study because I am an anaesthetist. The lowest acceptable inspired oxygen concentration during anaesthesia is 30%. I doubt whether there are many who would think it acceptable for me to undertake research by deliberately inducing hypoxia in patients in my care, even if they were capable of consenting.

No known mechanism links hypoxia and sudden infant death syndrome

EDITOR—Parkins et al's study prompts comment on its misinterpretation of published material, flaws in study design, and conjecture in relating the experimental observations to risk of the sudden infant death syndrome.1 Our study of arterial oxygen saturation in Tibetan and Han infants born in Lhasa, Tibet,2 was wrongly cited in support of the hypothesis that hypoxia increases the risk of apparent life threatening events and sudden death in infancy.

Our results related to subacute infantile mountain sickness, a form of pulmonary hypertension and right heart failure.3 This syndrome has a slow onset and clear pathological abnormalities; it is not similar to the sudden infant death syndrome.

Parkins et al's study is compromised by looking at a sample with several identifiable risk factors for abnormal oxygen transport, pulmonary function, and control of breathing. Characterisation of the outcome measure (saturation 80% for 1 minute) as “severe” and “prolonged” desaturation is a function of the testing procedureadministration of 15% oxygen at sea leveland has little relevance to physiology at high altitude.

Parkins et al's study is compromised by looking at a sample with several identifiable risk factors for abnormal oxygen transport, pulmonary function, and control of breathing. Characterisation of the outcome measure (saturation  80% for  1 minute) as “severe” and “prolonged” desaturation is a function of the testing procedure—administration of 15% oxygen at sea level—and has little relevance to physiology at high altitude.

Our observations at 3100 m in Leadville, Colorado, United States, have shown that healthy infants routinely experience arterial oxygen saturations <80% during sleep; some have a slower postnatal fall in pulmonary artery pressure compared with at sea level.4 Recent studies showed periodic breathing accompanied by fluctuating arterial oxygen saturation in infants who were 1 week to 3 months old.5 Four of 35 infants had repetitive desaturations <60-70% in the first 10 days of life. Two more developed pulmonary hypertension. There is no basis to equate these findings with the sudden infant death syndrome.

Finally, we question the assumptions that link airway hypoxia and the sudden infant death syndrome. What is the postulated mechanism that would cause death hours or days after return to sea level? The subjects were not monitored for hypoventilation, inequalities in ventilation:perfusion ratio, or reactive pulmonary hypertension in the time of interest (14-41 hours) after hypoxic challenge. The authors create a link between airway hypoxia and the sudden infant death syndrome through the example of hypoxaemic episodes that can complicate bronchiolitis. It is important not to blur the distinction between the sudden infant death syndrome and recognised clinical entities, such as pulmonary hypertension or bronchiolitis.

The exposure of infants to 15% oxygen yielded little information about physiology at high altitude or the sudden infant death syndrome. It did serve to create fear among parents planning air travel or visits to high altitude with young infants and to obscure understanding of the causes of infant mortality at high altitude worldwide.

Danger to babies from air travel must be small

EDITOR—Parkins et al's study has added to our understanding of the response to modest hypoxia in infancy.1 Unfortunately, however, media reports of the study have provoked public fear about the safety of air travel for young babies, and in particular their risk of subsequent sudden infant death syndrome.

As we were aware of the two cases of the sudden infant death syndrome that occurred after long haul air travel and were alluded to in Parkins et al's paper, we included questions about air travel in the previous month in the questionnaire used in the final year of the confidential inquiry into stillbirths and deaths in infancy and sudden unexpected death in infancy, 1995-6. The study covered a population of 17.7 million people, and 93% of parents of babies who had died of the sudden infant death syndrome participated. We have data on 130 cases of the syndrome and 528 controls. None of the babies who had died had flown in the month before—in fact, none had ever made an air journey—whereas two controls had. If the controls were representative of the general population the incidence of air travel would be 1 in 250 infants per year, so roughly 2400 infants would travel by air in the United Kingdom every year on non-domestic flights. If there is any danger to infants from air travel it must therefore be very small.

If we are ever in danger of forgetting the hazards of extrapolating clinical research findings from one situation to another we should remember that for many years mothers were advised to sleep their healthy term babies in the prone position, on the basis of physiological studies in preterm babies. The reversal of this advice, and the consequent reduction in the rate of the sudden infant death syndrome, remains a triumph of applied epidemiology over clinical science.

Study methods need to be appropriate

EDITOR—The methods Parkins et al used in their study were flawed and the reported data underanalysed and overinterpreted.1 Hypoxia causes a multisystem response that includes effects on arterial oxygen saturation, heart rate (cardiac output), breathing pattern and effective ventilation, and arousal state.

The arousal state modifies cardiorespiratory control and thus the response to hypoxia.2 These should have been measured and analysed before a conclusion could be drawn that responses were unpredictable and implied risk. Guidelines have been published for investigations that take account of these interrelations.3 These investigations are non-intrusive and different from those used by Parkins et al, which failed to prove the hypothesis.

Parkins et al contradict themselves in reporting that electrocardiography was used in their study, but reporting further down that it was not used. Only arterial oxygen saturation was used to assess hypoxic responsiveness, but it is sensitive to movement and two of six infants were erroneously withdrawn. The beat to beat mode does not improve accuracy of measurement but is more subject to motion artefact than averaging modes commonly used. Baseline data were reported only in relation to regular breathing, which was claimed to represent quiet sleep on the basis of a study that did not measure sleep.4 Periodic breathing normally occurs in quiet and active sleep, with arterial oxygen saturation cyclically below 90%, which increases with hypoxia and hyperthermia.2 Three of 20 healthy newborn infants responded to hypoxia with periodic breathing, lower transcutaneous oxygen tensions, and higher transcutaneous carbon dioxide tensions.5 Some infants as well as some adults thus normally respond this way when they are asleep. Is the blue bloater distinguishable from the pink puffer at birth?

Parkins et al analysed only 27% (confidence interval 0% to 53%), 16% (0% to 44%), and 25% (0% to 83%) of periods spent in regular breathing pattern before, during, and after hypoxic challenge. This is hardly a representative sample period or a safe monitoring method. An arterial oxygen saturation value <80% for >60 seconds is an arbitrary cut off and does not justify the clinical and alarming term severe hypoxaemia. We recently completed a study (table) during which we observed episodes of low arterial oxygen saturation in healthy infants during sleep while they were breathing air.

Measurements of arterial oxygen saturation vary. To have based such an important and emotionally charged study and conclusions, such as defining risk, on such inadequate methods and analysis is not acceptable. Investigators, funding agencies, and ethics committeesshould be aware of appropriate methodology.

Public must be warned of weak evidence for risk of serious harm

EDITOR—The public look to doctors for an evaluation of evidence and for advice. Parkins et al found that infants experience hypoxia when they are exposed to 15% oxygen.1 Milner commented in his editorial that all the epidemiological evidence indicates that flying is safe for healthy children in the first year of life.2 This claim is based on the reassurance given by a representative from British Airways, who said that the company would have investigated any incident.

There is some limited physiological evidence that a risk exists, but no epidemiological evidence. The public should be informed early of weak evidence that indicates a risk of serious harm but it should also be informed of the limitations of that evidence. This is consonant with public outrage at the withholding of early evidence that bovine spongiform encephalopathy presents a risk to humans.

More research is needed, but we also need discussion of how to evaluate the strength, quality, and relevance of evidence and how to translate these into practical advice for the public. Doctors, guided by experts in reviewing research, should judge whether advice along the following lines is justified.

“Some research suggests that there may be an increased chance of infants dying during and in the first few days after a long airline flight. However, this evidence is weak. It is not strong enough to make a recommendation that infants should not fly. Airline travel may be as safe for infants as it is for adults, but further research is necessary to clarify this. In the meantime, it would be prudent that if parents do choose to fly with a child under 1, they should be careful with sedatives and when children have colds.”

Even weak evidence can make a difference. My wife, who is an anaesthetist, and I recently flew with our 9 month old daughter. We sedated her with a low dose of trimeprazine (2 mg/kg), a commonly used sedative. At one point she stopped breathing and experienced altered consciousness for around 30 seconds. My wife was about to start oxygen treatment when the child improved with stimulation. We would not fly again with a child who is under 1 unless it was important. After reflection on Parkins et al's findings and our experience we would not use sedatives again. Like most people, we are averse to loss.

Risks associated with hypoxia during flights need to be investigated

EDITOR—Milner's editorial,1 which accompanies Parkins et al's study,2 states that flying is probably safe for infants. Because of birth events, however, not all people are born equal, and magnetic resonance imaging of children who develop seemingly normally after hypoxic-ischaemic encephalopathy at birth shows that their brains are not normal.2 Magnetic resonance imaging is likely to show similar morbidity in some children after seemingly uncomplicated deliveries and may provide crucial information in cot deaths. Immaturity of the blood-brain barrier and persisting oedema in the ischaemic penumbra in the midbrain may be adversely affected by the acute hypoxia in aircraft at cruising altitude, especially when exposure is maintained for many hours on long haul flights. The minimum oxygen content allowed in the workplace at sea level by the British Health and Safety Executive is 18%. Parkins et al calculated that the partial pressure of oxygen at the cabin “altitude” of commercial aircraft is equivalent to breathing 15% at sea level, but the work of breathing is much greater at sea level than at altitude because of the greater density of 15% oxygen in nitrogen. A small reduction in haemoglobin saturation reflects a much greater fall in the plasma oxygen tension, which alone determines the transport of oxygen into tissues. When haemoglobin saturation is reduced from 95% to 80%, the plasma oxygen tension falls from 12.6 kPa at sea level to about 6.6 kPa—a reduction of 47%. Hypoxia during flying may be associated with angina, stroke, and an increased risk of deep vein thrombosis.3 The symptoms may be delayed because these conditions may take hours or days to develop, which allows the relevance of the flight to be disputed.

The infants in the study were exposed to hypoxia for only six hours, but many flights last for over 12 hours. The current maximum cabin altitude for commercial aircraft of 2438 metres derives from the introduction of the jet engine and cabin pressurisation when flight durations were limited. Although living at high altitude is associated with an increased capillary density in the brain and other organs, this acclimatisation to hypoxia takes several weeks.

There is a trend in aircraft manufacture to use a higher cabin altitude,5 and flight durations of over 12 hours are common. The upper limit of 2438 metres, established many years ago, needs to be revised. The risks that are associated with the hypoxia that some newborn infants and adults experience while flying need to be investigated.

EDITOR—The sudden infant death syndrome is the commonest cause of infant death in developed countries. It is therefore important to understand the mechanisms responsible. West states correctly that our research was non-therapeutic. It did, however, conform to guidelines set by the local research ethics committee, and the risks involved were minimal.

We believe that most parents do not understand the hypoxic effects of air travel or visiting high altitude. Our previous research had shown a possible link between airway hypoxia, particularly during respiratory infection, and sudden infant death. We think that our approach is the most ethical and appropriate way of investigating the concerns we had raised.

We were aware that our cut off point of oxygen saturation (80% for 60 seconds) was the mean lowest value documented during sleep in healthy infants living at high altitude.1 Hypoxic challenge ended prematurely in a minority of our sample, which suggests that this response is uncommon in healthy infants living at sea level.

Johnson et al's measurements of oxygen saturation are 5-6% lower than our reference values; their observation that healthy infants can have saturation values of <80% is to be expected and does not contradict our protocol. The terminology in our article (“severe and prolonged desaturation”) related only to the context of our research design and did not reflect a clinical indication of concern. This may have influenced Pace to believe that clinically relevant hypoxaemia was present in four of our babies during the study. In the context of anaesthesia, where children are sedated and breathing additional inspired oxygen, saturations of 80% would of course indicate a clinical problem. In the context of our study they represent a variability in responses to a physiological challenge that babies undergo daily, either in air travel, living, or visiting at high altitude.

Niermeyer et al did not report on apparent life threatening events.1 They described cyanotic episodes during feeding and sleep, and commented on deaths from airway hypoxia. Heath reported on 15 infants who died 2.1 months after moving to high altitude.2 Data from Sui et al3 are also cited by Niermeyer and Moore. Increased muscle in pulmonary arterioles has been reported in infants who died of the sudden infant death syndrome.4 Bronchiolitis is commonly identified at postmortem examination in infants who have died suddenly and unexpectedly.5

We are unaware of data that suggest that our experimental groups—healthy infants and siblings of victims of the sudden infant death syndrome—are at increased risk of “abnormal oxygen transport, pulmonary function, and control of breathing.” Mechanisms for airway hypoxia that trigger sudden death in infants potentially include changes in the pulmonary vasculature and the smooth muscle in the peripheral airways of the lung. We were careful not to imply that we had found an explanation for the syndrome.

Pace thinks that our study should have been conducted on an aircraft. This incorrectly assumes that we were primarily interested in the effects of air travel. We agree with Platt et al that provocation in the media of fear about the safety of air travel for babies was inappropriate and premature. We appreciate their epidemiological data but question whether sample size was sufficient to discount a relation between air travel and sudden infant death. We have now received letters from parents and doctors that identify 10 infants who have died within a few days after an airline flight. We were also informed about four infants who had experienced apparent life threatening events, one of whom required cardiopulmonary resuscitation on the plane. One of these died shortly after the plane landed.

Johnson et al think that our study should have conformed to published guidelines before we concluded that responses to hypoxia were unpredictable. Their own guidelines do not, however, seem relevant to our model. They have misunderstood our analysis, which was not restricted to periods of regular pattern breathing.

We agree with Savulescu that research should be publicly available. We aimed to elucidate responses to airway hypoxia in infants who normally breathe 21% oxygen. Our findings and the subsequent correspondence justify further investigation. Safety of air travel safety for babies can probably best be determined by observations during and after flights.