Respiratory Physiology & Neurobiology

Supramedullary influences on cough

John Widdicombea, Ron Ecclesb, and Giovanni Fontanac

University of London, 116 Pepys Road, London SW20 8NY, UK
Common Cold and Healthcare Clinical Trials, Cardiff University, Wales, UK
Sezione Di Fisiopath. Resp., Universita Di Firenze, Italy

Accepted 28 February 2006. Available online 18 April 2006.


The evidence for supramedullary influences on cough is largely indirect. Cough can be voluntarily induced or inhibited, functions usually thought to reside in the cerebral cortex. A sensation of ?urge-to-cough? usually precedes cough due to an airway irritant stimulus, and this may well involve the cerebral cortex. In conditions with interruption of the pathways between the cortex and the brainstem, such as strokes and Parkinson's disease, voluntary cough may be inhibited without disruption of reflex cough from the larynx or lower airways. ?Habit cough?, like Tourette's syndrome, is assumed to be cortically mediated. Placebos and many treatments based on complementary medicine are effective in inhibiting clinical cough, and the site of action is likely to be the cerebral cortex. In sleep and in anaesthesia cough is depressed and, again, this seems likely to be at a cortical level. However there are few or no experimental or clinical observation as to the localization and functions of supramedullary areas responsible for cough. It is a field of research wide open for exploration.


We can both induce and inhibit cough voluntarily (Hutchings et al., 1993a and Lee et al., 2002a), and it is reasonable to suppose that these controls are located in, or possibly just below, the cerebral cortex. However there is little or no hard evidence about the exact sites involved. Sleep and general anaesthesia inhibit cough (Hsu et al., 1994, Nishino et al., 1996, Widdicombe and Singh, 2006, Coyle et al., 2005a and Coyle et al., 2006) and we can assume, in the absence of evidence, that these actions are at cortical level. Similarly, in comatose patients the cough reflex may be markedly depressed, but reduced consciousness is not the only factor in cough depression (Moulton and Pennycook, 1994). Of note, the cause leading to coma, e.g. drug abuse, trauma or post-ictal state, appears to be an independent contributing factor: for a given level of reduced consciousness, pharmacological compared with other causes of coma are associated with a more profound depression of the cough reflex (Moulton and Pennycook, 1994). An involvement of higher brain centres in cough control is also indirectly suggested by the fact that placebos (Eccles, 2006) and many treatments used in complementary (alternative) medicine (see below) can inhibit cough, even when administered in such a way as to eliminate physical or physiological inputs. The best direct evidence for cortical or subcortical influences on cough control comes from patients with neurological lesions due, for example, to strokes (Fontana and Widdicombe, 2004) or Parkinson's disease (Fontana et al., 1998 and Ebihara et al., 2003); here voluntary cough is often seriously impaired or even lost, and it is reasonable to suppose that nervous pathways joining the cortex or subcortical areas to the brainstem have been disrupted.

In man, the lack of experimental evidence mainly stems from the obvious difficulty in performing appropriate experiments; in general one can only use the natural experimental material offered by disease, and here precision is usually lacking. With experimental animals, much used to study cough, voluntary cough is impossible to evoke. However an interesting study by Pinto et al. (1995) showed that guinea-pigs could be conditioned to cough by exposing them to capsaicin and camphor together, although only the former agent given alone caused cough; 4 days after the capsaicin coughing, exposure to camphor alone caused cough, presumably by ?associated learning?. In animals cough-like responses can be induced or inhibited by stimulation or removal of structures above the brainstem, but the applicability of these results to humans is uncertain. Furthermore the animals are nearly always anaesthetized, which would inhibit or block the supramedullary activation of cough. Modern methods such as transcranial magnetic stimulation (TCMS) and scanning tomography (Fong et al., 2004) show promise but have yet to be applied to study cough.

Reflex cough can adopt various stereotypical patterns: the classical definition is a deep inspiration (inspiratory phase), followed by a forced expiration against a glottis which is initially closed (compressive phase) and then opens to allow the expulsive phase of cough (Leith, 1977, Korpas and Tomori, 1979 and Widdicombe and Fontana, 2006). This single and typical cough event may be followed by further coughs and/or expiratory efforts without intervening inspirations?a cough bout or epoch. If food or an irritant is inhaled into the larynx there is an initial expiratory effort (the expiration reflex) with closing and then opening of the glottis, possibly followed by further expiratory efforts and coughs (Korpas and Jakus, 2000 and Widdicombe and Fontana, 2006). Although reflex clinical cough can be suppressed (Bucher, 1958 and Hutchings et al., 1993a; see below), there seems to be little voluntary control over its pattern; it is not known whether the expiration reflex can also be suppressed, but this may be unlikely in view of its usually violent nature. In contrast, voluntary cough may take almost any form the cougher wishes, and the phases of cough, inspiratory, compressive and expulsive, can be largely controlled independently.

Voluntary control of cough

Induction of cough
It is assumed that voluntary cough is initiated in the cerebral cortex, a plausible view in the light of the known functions of the cortex, but one for which there is little experimental evidence. Electrical stimulation of the cortex in unanaesthetized humans and anaesthetized members of other species has never been reported to cause clear-cut cough (Foerster, 1936, Penfield and Rasmussen, 1949 and Hast et al., 1974), although cough-like patterns have been described on stimulation of the cortex in anaesthetized cats. For instance, it has been shown (Kase et al., 1984) that electrical stimulation of the suprasylvian gyrus, a region possibly related to laryngeal muscle control in monkeys (Simonyan and Jurgens, 2005), can initiate cough. Conversely, stimulation of the anterior cingulate gyrus suppresses the cough evoked by superior laryngeal nerve (SLN) stimulation (Kase et al., 1984). Laryngeal adductor muscle contraction, vocalization and expiratory efforts are seen when the supplementary motor area within the sylvian fissure is stimulated or activated by TCMS (Fong et al., 2004). This area has motor output to the nucleus ambiguus in the brainstem, the motor nucleus for the laryngeal muscles. Unlike the cerebral cortex, the medulla when stimulated can induce cough and/or expiratory efforts (Borison, 1948), although these have not been shown to have the full characteristic pattern of the cough and expiration reflexes. In man, TCMS studies have led to the identification of areas of neural activation by volitional inspirations (Evans et al., 1999). These areas include the superior motor cortex, the premotor cortex, the supplementary motor area, the infero-lateral sensorimotor cortex, the prefrontal cortex and the striatum. Whether these areas participate also in the production of cough motor patterns has not been established, but seems plausible.

Voluntary cough may also be related to the sensation ?urge-to-cough?.
Two published studies have shown that with activation of reflex inputs to cough an urge-to-cough has a lower threshold than the cough itself. Davenport et al. (2002) gave unanaesthetized humans different concentrations of capsaicin aerosol and assessed the strength of the sensation urge-to-cough and the cough itself, and found that the latter required a higher concentration. Fontana and colleagues (unpublished) obtained similar results with capsaicin and distilled water aerosols. Similarly, Paintal (1995) and Raj et al. (1995) stimulated bronchopulmonary receptors with intravenous lobeline to produce cough, and found that doses too small to excite cough did cause urge-to-cough. Presumably the subjects felt a stimulus to cough but neglected it, just as the sensation to itch does not always lead to scratching. With weak stimuli the urge-to-cough may or may not lead to cough, although it always precedes it; with strong reflex coughs due, for example, to inhaled irritants, cough is inevitable and cannot be suppressed. These studies led to the concept that reflexly-induced cough in humans is in part a behavioural response, or has a behavioural element, based on the cortex and superimposed on the reflex cough. Raj et al. (2005) tested this possibility by showing that in anaesthetized and intubated patients, and therefore with urge-to-cough absent, lobeline still caused cough or cough-like responses, and concluded that the behavioural element was weak or absent; however this does not invalidate the existence of a behavioural urge-to-cough initiating or enhancing cough itself in conscious subjects. In addition since the patients were intubated there is some doubt whether they exhibited true coughs or the deep breaths known to be activated by lobeline.

Some of the strongest evidence for a diencephalic initiation of coughs comes from studies of patients with stroke (Addington et al., 1999, Hammond et al., 2001 and Widdicombe and Singh, 2006) or Parkinson's disease (Nakashima et al., 1997 and Fontana et al., 1998); they often have a weak or absent cough, both reflex and voluntary, a condition that may lead to aspiration pneumonia. Presumably the subcortical lesions block motor pathways for voluntary cough from the cortex. Although Fontana et al. (1998) showed that the cough threshold to inhalation of nebulized distilled water was unaltered in Parkinsonism, suggesting that sensory pathways were not affected, later work by Ebihara et al. (2003) showed that in advanced Parkinsonism the citric acid threshold was heightened, although there is no convincing reason to believe that this depression was at a supramedullary level, it could be due to interruption of cortical pathways facilitatory to cough.

In another study on patients with stroke, the site of the lesion was related to voluntary cough (Stephens et al., 2003). Most of the patients with left middle cerebral artery infarcts had loss or weakness of voluntary cough, whereas none of those with right-sided infarcts had similar loss of cough. No left-handed subjects were studied. The results suggest that the left cerebral hemisphere is dominant in the voluntary control of cough in right-handed subjects. The loss of cough in patients with stroke has led to the concept of ?brainstem shock? (Addington et al., 1999 and Addington et al., 2005), whereby general depression of brainstem function may inhibit reflex or voluntary cough or both. This idea has not been established, but if it is correct then the loss of cough might be beneficial, since cough elevates cerebrospinal fluid pressure and might exacerbate the brain injury.

Unlike patients with stroke or Parkinsonism, those with ?locked-in syndrome? lack a voluntary cough just as they lack voluntary control of their breathing, but have a normal cough reflex from the trachea (Heywood et al., 1996). The lesion in this condition is thought to be in the pons, and presumably it interrupts the voluntary cough pathways from the cortex. In contrast to the locked-in syndrome, patients with congenital central hypoventilation syndrome (CCHS) have ?normal? voluntary control of respiratory muscles, including cough, but brainstem function is defective, as shown by weak or absent chemical control of breathing. An early study (Shea et al., 1993) suggested that reflex cough was also deficient in CCHS, possibly due to the removal of cortical facilitatory pathways for cough; however more recent work by Fontana (Widdicombe and Fontana, 2006; and unpublished) showed that patients with CCHS do show a reflex cough in response to inhalation of distilled water.

A final clinical condition relevant to the cortical induction of cough is ?habit cough? (cf. Tourette's syndrome) (Cohlan and Stone, 1984 and Tan et al., 2004). In view of its absence in sleep and general anaesthesia, and its depression by hypnosis and other approaches of complementary medicine (see below), it seems reasonable to suppose that the cough has a cortical origin.

Fig. 1 illustrates in general the links between the cerebral cortex and the brainstem in the control of cough. In summary, both on grounds of the known physiology of the central nervous system, and on the interpretation of cough defects in a number of central nervous diseases and conditions, it is likely that the mechanism(s) involved in the voluntary induction of cough reside(s) in or close to the cerebral cortex. Beyond that there is little indication of its precise location and functional properties.

Fig. 1. Cough model to illustrate reflex and voluntary control mechanisms. Irritation of airway receptors may cause reflex cough via a brainstem cough control area. A sensation of irritation may cause cough via higher centres such as the cerebral cortex. Cough may be voluntarily initiated and inhibited via the cerebral cortex that influences cough by two pathways; via the brainstem, and via a descending pathway to the spinal cord. (modified from Eccles, 2003).

Inhibition of cough

Hutchings et al. (1993b) showed that the frequency of cough due to inhalation of capsaicin aerosol could be inhibited by 90% in healthy subjects (Fig. 2). Lee et al. (2002a) found that clinical cough due to upper respiratory tract infection can be suppressed for 3?20 min. They concluded that the mechanism of cough suppression was similar to that of the placebo effect (see later), and it seems reasonable to suppose that the voluntary inhibition arises at or close to the cerebral cortex. Interestingly, the cough suppression in the patients was related to psychological factors, the ability being greater in patients with obsessional symptoms and mood. There may be an analogy with habit cough in patients with Tourette-like syndrome, who can suppress their cough, if unwillingly, when told to do so (Cohlan and Stone, 1984 and Tan et al., 2004).

Fig. 2. Mean number of coughs per 30 s with standard error bars from 24 subjects produced by inhalation of capsaicin. Round symbols indicate coughs without any voluntary suppression (from Hutchings et al., 1993a).

The placebo suppression of cough could be due to the release of endogenous opioids in the brain (see later), and it has been suggested that the same may be the basis of voluntary suppression. This may seem unlikely in view of the rapidity of voluntary cough-inhibition. The possibility has been tested by the use of opioid antagonists. Hutchings and Eccles (1994) found that neither the opioid antagonist naltrexone nor the opioid agonist codeine, compared with placebo, changed the degree of voluntary suppression of cough due to capsaicin.

All studies have measured the frequency of cough, and it is not known to what extent the intensity of clinical or induced cough, compared with voluntary cough, can be voluntarily controlled, as is presumably the case. Also there seem to have been no studies on the voluntary inhibition of the expiration reflex due to laryngeal irritation.

The placebo effect

This is dealt with in detail elsewhere in this Special Issue (Eccles, 2006; see also Eccles, 2002 and Eccles, 2003), and here we will only summarize evidence that placebos work at a cortical or close subcortical level. For most over-the-counter treatments of cough, about 85% of their effectiveness is due to placebo action (Pavesi et al., 2001 and Schroeder and Fahey, 2002). Eccles has suggested that the placebo effect has two components: a ?physiological? action due to the adjuncts (such as sweetness or flavour) to the true antitussive agent in the therapy (Eccles, 2006); and a ?true? placebo action (Lee et al., 2005).

The physiological placebo effect

The analysis of the physiological effect is in its early stages. Examples may be the sweet taste of sugary ingredients, the soothing of sensation due to emulcients and the cough-inhibiting result of excitation of other sensory inputs, e.g. the cold sensation associated with menthol (Eccles, 1994). At present it is impossible to say if these effects work at least in part at a brainstem level, by inhibiting cough-generating areas here.

The problem is partly semantic. If a mixture of a pharmacological antitussive drug such as codeine contains, for example, menthol, which is known to inhibit cough by physiological pathways, is it reasonable to distinguish them as ?pharmacological? and ?physiological?, since both are both; and should we put the ?physiological? into the placebo category? But the terms have been clearly defined (Eccles, 2006), and we believe that the distinction is helpful and should encourage future research.

Animal experiments, for example with decerebrate animals in which cortical function is eliminated, may be needed to test the site(s) of action of the ?physiological? inputs. However, it seems worthwhile recalling that a ?physiological? respiratory effect by the antitussive agent l-menthol seems likely in the view of its known action at the level of laryngeal and nasal cold-sensitive endings (Sant?Ambrogio et al., 1991). In anaesthetized animals breathing through a tracheostomy, activation of laryngeal cold-sensitive endings by constant airflows of either cold air or warm air with the addition of l-menthol, delivered through the isolated upper airway in the expiratory direction, greatly reduced ventilation (Orani et al., 1991 and Sant?Ambrogio et al., 1992). This depressant respiratory response could be completely abolished by cutting the superior laryngeal nerves and by nasal anaesthesia (Orani et al., 1991 and Sant?Ambrogio et al., 1992). Ventilatory depression by upper airway cooling has also been reported in conscious humans (McBride and Whitelaw, 1981 and Eccles and Tolley, 1987). Taken together, the results of animal and human studies point to a significant antitussive physiological effect by l-menthol. Since it seems well established that coughing involves many of the same brainstem neurones as those subserving the eupnoeic pattern of breathing (Bolser and Davenport, 2002 and Shannon et al., 2004), inhibition of the respiratory neural network may contribute to explain the cough-sedating activity of l-menthol.

However it is relevant that many afferent inputs to the brainstem have been shown to inhibit the cough reflex at brainstem level, e.g. those from pulmonary, cardiac and abdominal C-fibre sensory receptors (Hanacek et al., 2006: this Special Issue). What is important is that in the examples of physiological inhibition of cough given above, there is always sensation involved, which implies that cortical and close subcortical areas must be activated. Just as afferent inputs from airway ?cough receptors? can evoke both brainstem reflex cough and cortical urge-to-cough, the latter presumably augmenting the reflex cough, so it is possible that the physiological inputs to cough associated with many placebos can act at both levels.

The ?true? placebo effect

The true placebo effect, seem for example with tablet- and capsule-administered antitussive agents, is thought to be related to psychological factors such as belief in the effectiveness of the treatment and the patient's attitude to the therapist (Lee et al., 2002b and Eccles, 2006). Presumably these processes take place at a cortical or near-subcortical level.

The mechanisms of the presumed cortical action of placebos have not been studied for cough. But if they are similar to those for analgesics, they may involve local release of opioids. This possibility does not seem to have been tested. Although, as mentioned above, opioids do not seem to be involved in the voluntary suppression of cough (Hutchings and Eccles, 1994), this may not rule out a role for them in the placebo effect.

There is a vast literature on complementary medicines and cough, in particular relating to herbal medicine. It is fair to assume that in general only positive results get published.

Thus very many herbal treatments seem to be effective, but it is hard to find an adequate placebo-controlled study. Some studies have used an established antitussive agent (usually codeine) as a control, but this tells us little about a placebo effect and, if there were one, whether the effect would be ?physiological? or ?true?. Only in the latter instance might a cortical mechanism be implicated. One might expect a large ?physiological? effect with many herbal treatments.

Some complementary treatments for cough seem certain to act at a cortical level. These include hypnosis (Anbar, 2002 and Anbar and Hall, 2004) and psychotherapy (Gay et al., 1987, Blager et al., 1988, Lokshin et al., 1991 and Cowan et al., 2001). Acupuncture can also be effective (Chen et al., 2002), but there is no scientific indication as to where it works in the nervous system.

Inhibition of cough in sleep

Anecdotally it is well established that cough is inhibited in sleep, but there have been few published measurements. And the results from these studies do not establish that the inhibition is at a cortical level, although this seems a very reasonable assumption. Hsu et al. (1994) did 24 h recording of cough in asthmatics and other patients with chronic cough, and found that in deepest sleep cough was reduced to 10% of that for the rest of the day. Coyle et al. (2005a), Coyle et al. (2006) obtained similar results, as did Power et al. (1984) with patients with chronic bronchitis and emphysema. Patients with habit cough (c.f. Tourette's syndrome) do not cough when asleep (Weinberg, 1980, Cohlan and Stone, 1984 and Tan et al., 2004) and, since their daytime cough is psychogenic, the cerebral cortex is almost certainly its site of origin. Their cough is not affected by antitussive agents (Schuper et al., 1983).

There have been a few studies on induced cough during sleep. Jamal et al. (1983) showed that the citric acid threshold for cough was increased during human sleep. This confirmed the classic work by Sullivan et al., 1978 and Sullivan et al., 1979 with sleeping dogs; they showed that during sleep distilled water in the larynx or trachea was a far weaker stimulus to cough, and that the cough only occurred after arousal (i.e. in a sense only after the dogs were awakened). The expiration reflex from the larynx was also depressed during sleep, but could be elicited without arousal; the protective advantage of this against aspiration is clear. Similar results were obtained with sleeping cats by Anderson et al. (1996).

Sleep is presumably acting at a cortical level. The suppression of reflex cough could be at a brainstem level, although sleep seems to have little action on neural and chemical control of breathing. If the effects of sleep are at a cortical level, this must be due to a new balance between the inhibitory and excitatory influences on cough from that site, with the former predominating, but this is a rather imprecise conclusion.

General anaesthesia and cough

General anaesthesia depresses cough, both in experimental animals and in humans. This can occur with levels of anaesthesia that do not depress tidal volume and frequency or their chemical control (Nishino et al., 1988 and Tagaito et al., 1998). The similarities with sleep are clear.

The depression depends on the type and depth of the anaesthesia (Nishino et al., 1988 and Tagaito et al., 1998). Speaking generally, as anaesthesia deepens, cough from the larynx is first suppressed and then the expiration reflex, at a time when the glottal closure and apnoeic reflexes may be enhanced; the last in particular is resistant to anaesthesia. Anaesthesia may depress the intensity but not the frequency of induced cough (Tatar and Korpas, 1983), and in humans the expiratory but not the inspiratory force of coughing (Nishino and Honda, 1986). Light volatile anaesthetics are less potent in suppressing cough than are non-volatile ones; while nitrous oxide, at least in cats, does not inhibit cough at all (Tatar and Korpas, 1983).

In view of the complexity of the actions of anaesthetics on coughing, it is not surprising that no firm conclusions can be drawn as to the site or sites of action, although they are generally assumed to be cortical or just subcortical. The fact that anaesthetics both block cough and cause sleep, and the similarities between the respiratory changes in sleep and anaesthesia, may be the main support for this conclusion.


Cough is a complex phenomenon often implicating signals arising from supramedullary brain areas. Little is known about the cortical motor mechanisms brought into action when one tries to suppress cough or about the mechanisms subserving voluntary cough. The motor pathways between cortex and brainstem that could facilitate or inhibit cough have not been investigated. The relationships between respiratory sensations and the cough reflex are ill defined. Exploration of the mechanisms might have an important impact on research of antitussive drugs.