Research in children and ethical limits: the water laryngospasm challenge


The medical profession has a duty, in accordance with the Helsinki Declaration, to protect the rights of participants in clinical research. If an experiment is deemed high-risk, with a significantly greater probability of harm than that observed during standard care, then it challenges the boundaries of this ethical responsibility. The non-randomised interventional study by Erb et al., in this issue of Anaesthesia [1], could be accused of posing such a challenge, particularly as it involves children. The investigators subjected forty children undergoing minor surgery to repeated water challenges of the larynx to induce laryngospasm. The aim was to investigate the benefits of intravenous lidocaine by giving it to all patients and recording the ease of provoking laryngospasm over a ten-minute period. One might wonder whether there is any benefit an individual patient could possibly gain from such a study, when its goal is to induce what is usually a complication that anaesthetists strive to avoid. Does this study tell us something so important that it justifies the increased risk?

The science

The first issue to address is the reliability and validity of eliciting airway protective reflexes with topical water. Laryngeal sensory receptors respond to various stimuli including air flow/pressure, touch and chemical irritation that may result in apnoea, cough, laryngospasm and if severe, arterial desaturation and bradycardia. Protective airway reflexes are powerfully modulated by complex and poorly understood interactions with central respiratory receptors and other feedback loops, including pulmonary stretch receptors [2]. Simple airway rescue manoeuvres required to overcome laryngospasm, such as jaw thrust, manual lung inflation and provision of oxygen, may influence the laryngeal reflex. In children, bradycardia following laryngospasm may be pronounced if also associated with hypoxia [3]. Hypercarbia secondary to hypoventilation may even overcome the apnoeic reflex by overwhelming central respiratory stimulation. Thus, there are many opposing confounding factors requiring consideration. Most importantly, there are considerable maturational changes in both laryngeal receptor sensitivity and central brain respiratory control with age. For example, neonates demonstrate prolonged apnoea long after the cessation of the primary laryngeal irritant, yet in adults, ventilation may be preserved despite repeated laryngeal stimulation [4].

It is hard to see how Erb et al. could control for these factors since even a brief airway rescue procedure for laryngospasm can modify the airway protective reflex, as described above. Furthermore, animal studies suggest that repeated laryngeal stimulation over a short time (minutes) may progressively dampen the reflex’s potency [5, 6]. A reduction in the rate of laryngospasm over ten minutes, as observed by Erb et al., may be due to adaptive dampening of the reflex rather than a true lidocaine effect. Without a control arm and adjustment of confounding variables, it is impossible to interpret repeated laryngeal responses to water stimulation as the normal response of the human larynx across age groups has not yet been adequately studied or defined. Those animal studies that have tried to produce a reliable model of laryngospasm remain small and largely experimental [4]. Both the anatomical distribution of chemical and mechanical (tactile) receptors in the larynx, and the type and depth of stimulus required to induce laryngospasm, vary. Although the majority of laryngeal sensory input travels via the superior laryngeal nerve, different irritants may produce different protective reflexes of different intensity [3–5, 7].

In the few paediatric studies where water was used to elicit vocal cord closure, a wide variation in laryngospasm rates (from 25% to 90%) has been observed, despite similar methodology and technique [8–10]. In fact, 90% (169/189) of all children exposed to this technique were reported by the same research group [1, 8, 9]. Whilst various explanations for these differences may be offered, such as different depths of anaesthesia or induction agents, the methodology adopted has usually precluded interpretation of results by not adjusting for confounding factors.

Finally, as elegant and appealing as water stimulation of the larynx may seem, water is not usually found near the airway during normal anaesthesia; mechanical irritation of the larynx, for example through the use of a supraglottic airway, is more likely.

The ethics

There is no question that deliberately inducing laryngospasm exposes the patient to unnecessary additional risk, despite the authors’ claim to the contrary. Laryngospasm may be complicated by hypoxia, hypoventilation, gastric aspiration, pulmonary oedema and rarely, death; it is the commonest respiratory cause of cardiac arrest during anaesthesia [11]. In Erb et al.’s study, laryngospasm was induced successfully in up to 40% of patients on the first of the three water challenges. In a recent study of 1000 children undergoing anaesthesia for elective surgery [12], the incidence of laryngospasm was 4.4%; thus the risk of laryngospasm (and its potential complications) was increased almost tenfold. Despite expertly applied and timely rescue manoeuvres to treat laryngospasm, one patient required neuromuscular blockade and in one there was desaturation to 66%. All previously reported paediatric studies using this technique, mostly by the same group, report a higher complication rate, and in one publication a requirement for emergency neuromuscular blockade of over 50% [10]. Inducing laryngospasm in every patient according to the research protocol used cannot carry the same risk of general anaesthesia when the standard rate is usually well below 10%.

Is there a safer way to determine whether intravenous lidocaine is effective in reducing the incidence of laryngospasm? A randomised control trial (RCT) is the gold standard method as it adjusts for confounding factors and limits bias. A prerequisite is that there is equipoise regarding the use of lidocaine. There is good evidence of this, as there are no national or international guidelines mandating its use for preventing laryngospasm. Furthermore, there appear to be large variations in anaesthetic practice concerning even topical lidocaine, confirming that equipoise exists [12]. A sample size of over 1000 would be required to power a RCT with a baseline rate of laryngospasm of around 4–5%. Trials of this size have been successfully carried out in the paediatric critical care setting; not only is it feasible, but a multicentre RCT may provide answers that carry sufficient weight to shape future anaesthetic practice, albeit requiring great effort and significant resources. It is unquestionably a safer option than performing high-risk, low-yield research on a few patients, where failure to adjust for confounding variables limits interpretation of the results.

One final limitation of Erb et al.’s study is selection bias, with 63% of parents declining to give consent even with a ‘guarantee’ of safety offered by the investigators. This is a consistent feature of all other paediatric studies involving water-induced laryngospasm. It is one of the highest rates of parental refusal of consent in paediatric anaesthetic research and therefore begs the question: is it really acceptable to continue using children in potentially high-risk research, with limited scientific knowledge gained, if safer alternatives exist? In this modern age of governance and ethical responsibility, it is timely to address this issue.

Competing interests

No external funding and no competing interests declared.