Problems relating to clinical studies of cough

Studies that include cough as an outcome need to be interpreted in light of inherent methodological problems. First, although the history of cough is a commonly used variable in epidemiology,2 3the reporting of cough has poor repeatability. The chance corrected agreement (Cohen’s κ) between two occasions when cough was reported, is unacceptably low (0.14−0.19).4 5 In contrast, wheeze related questions have excellent κ values of 0.76−0.88.4 5 Second, cough, especially nocturnal cough, is unreliably reported compared with objective measures.6 7 In adults, the reporting and scoring of cough is dependent on psychological factors.8 Third, although cough scores using diary cards are widely used as an outcome measure, until recently no validated subjective cough scoring system was available.9 Compared to objective measurements, diary cards have been shown to be inaccurate for nocturnal cough6 and for metered dose inhaler use. Moreover, diary cards in studies can be parent completed6 or child completed, and these are not interchangeable.9 The accuracy of findings based entirely on diary cards is questionable.

Finally, a difficulty in the evaluation of the benefits of a therapeutic trial for cough is that cough usually resolves spontaneously (period effect). The widely quoted studies on “cough variant asthma (CVA)” were based on anecdotal evidence or were not placebo controlled.10-13 In trials that included a placebo arm, cough medications were no more effective than placebo.14 15 Guidelines that recommend a therapeutic trial of asthma medications for cough (and thus diagnose CVA) often fail to alert clinicians to this point.16

Traditional clinical model of cough and asthma

Based on the simple observations that some children with wheeze also cough, and that bronchoprovocation agents often stimulate both bronchoconstriction and cough in susceptible individuals, have led to the assumption that cough is part and parcel of the same mechanism as bronchoconstriction—the traditional clinical model. It has partly contributed to the establishment of common epidemiological and clinical practices.

In epidemiology studies, cough as a sole symptom has been used by some as synonymous with the diagnosis of asthma. Children with cough alone were categorised as “asthmatic”.2 17 18 Furthermore, the use of the symptom of cough alone as representative of asthma has partly contributed to the increased prevalence of asthma.18 19 In a paper that provided a breakdown of the method of how asthma was diagnosed, cough alone accounted for the largest rise in asthma prevalence.18 Between 1990 and 1993, the change in the cumulative prevalence of wheeze was 0.6% and that of troublesome cough was 3%.18

Kelly and colleagues showed that doctors were twice as likely to diagnose asthma on the basis of cough alone in 1993 compared with 1991 (22.6% and 10%, respectively).19 Of the children with no symptoms in the previous 12 months, 52% had a cough at some time that had been diagnosed as asthma.19 Although the prevalence of respiratory symptoms was unchanged, the number of children who were diagnosed with asthma significantly increased.19

Clinically, the advocation of the diagnosis and treatment of the symptom of cough alone as asthma16 20 is also based on the traditional clinical model. Several groups have advocated the use of inhaled or oral steroids in children with cough alone. In one protocol children with cough were presumed to have asthma and some children were given repeated courses of oral steroids.20

Cough is used as a marker of asthma stability in several asthma management guidelines21 and for assessing asthma severity.22 Guidelines for asthma management in children that state simply that cough orwheeze/dyspnoea are symptoms of asthma instability and therefore an indication to increase medications, are open to misinterpretation.21 In a functional severity score for asthma in children,22 a child with cough alone without other evidence of airway obstruction can be categorised as the second most severe group. In asthma guidelines the inclusion of the symptom of cough alone without qualification can lead to overtreatment.

Although there is no doubt that children with asthma can present with cough12 and, in some, cough is a major symptom in asthma exacerbations, there is no evidence that the same mechanism is responsible for cough and bronchoconstriction. Indeed the converse is true—the alternative model.

Alternative model

There is good laboratory and clinical evidence that the pathways for cough and bronchoconstriction are distinctly different. In the laboratory, bronchoconstriction and cough pathways can be separately inhibited. Medications (lignocaine, oral codeine) that inhibit cough have no effect on bronchoconstriction. Conversely, medications (cromoglygate, atropine) that inhibit the pharmacologically induced bronchoconstriction have no effect on the cough response in adults.23 24 Atropine and cromoglygate inhibit bronchoconstriction but have no effect on the cough response, while inhaled lignocaine abolishes cough but does not influence bronchoconstriction.23 24

Fujimura and colleagues examined the effect of airway calibre changes on cough receptor sensitivity (CRS) and found that methacholine and procaterol affected airway calibre, however neither affected CRS in adults with airway hyperresponsiveness (AHR).25 Similarly in children with stable asthma no correlation was found between CRS and airway calibre.26

Clinically, three groups have described a lack of correlation between cough severity and asthma severity.7 27 28 During the recovery phase of an acute asthma exacerbation, cough frequency objectively measured had no correlation with airway calibre measured by diurnal peak expiratory flow (PEF) or its variability.28In adults with stable asthma, cough frequency measured over 24 hours did not relate to asthma severity defined by FEV₁ (forced expiratory volume in one second) or diurnal variation in PEF.7 Nocturnal cough measured objectively in non-hospitalised children with asthma had no correlation with AHR, PEF variability, degree of morning dip in PEF, mean overnight saturation or treatment of asthma.27

In children with asthma, there was no relation between CRS and airway calibre (FEV₁) in both the acute and interval phase of asthma.29 In contrast, in the recovery phase of acute severe asthma, improvements in FEV₁ correlated with the reduction in sputum eosinophils and eosinophil cationic protein.30

Thus, while the trigger for wheeze and tachypnoea may be similar to that for cough, the pathways are distinctly different. Treatment with increasing doses of medications used for asthma of cough alone without any evidence of airway obstruction is questionable. Why the cough pathway is so easily stimulated in some is not known. The development of heightened CRS may account for the tendency of some patients to have troublesome cough.

Cough receptor sensitivity and cough

Cough can be mechanically or chemically stimulated endogenously (by airway secretions, mediators of inflammation) or exogenously by a variety of irritants (aspirated foreign materials, smoke, etc), and tussive agents (capsaicin, citric acid). Cough sensitivity and pattern depends on the site and type of stimulation.31 The more proximal airways (larynx to trachea) are extremely sensitive to mechanical stimulation. The more distal airways are more chemosensitive and less mechanosensitive.

In both children and adults, heightened CRS occurs in those with a dry cough but not in those with productive cough.32 33Prospective studies have shown an association between a decrease in symptoms and CRS of adults whose cough was secondary to angiotensin converting enzyme inhibitor use34 and upper respiratory tract infection.35

In children, a temporal association between increased CRS and cough has also been demonstrated.36 Some children with recurrent cough had increased CRS and others had AHR during the coughing period, which returned to control values when cough free.36Children with asthma who have cough as a major symptom of their acute episodes had a transient increase in CRS during the acute asthma exacerbation. In contrast, children who do not usually cough during their asthma exacerbations had no change in CRS during the acute and non-acute asthma periods.29 CRS of children with asthma during the non-acute phase were similar to controls.33

The pathophysiology as to why increased CRS develops in some is not known. Possible factors include increased calcitonin gene related peptide nerve density,37 release of tachykinins,31 persistent inflammation,38 and qualitative or quantitative surfactant abnormality.39Perhaps sensitivity of the cough receptors can be viewed in a similar way to sensitivity of the bronchoconstrictive pathway in susceptible people under certain circumstances.

Cough: is it asthma?

Several epidemiological studies also support McKenzie’s article1 that highlighted the problem of overdiagnosis of asthma based on the symptom of cough alone. In a prospective community study, 56% of children with recurrent cough aged 4−7 years, later became asymptomatic: 37% reported continuing cough and 7.2% developed wheeze.27 The proportion of children in the group who subsequently developed wheeze was similar to the asymptomatic group who developed wheeze on follow up (10%).27

Without intervention, recurrent respiratory symptoms in 8−10 year olds generally improve.3 In the absence of wheeze or shortness of breath, cough was not indicative of hidden or atypical asthma in most children with persistent nocturnal cough.40Wright and colleagues in their prospective study of infants followed up to 11 years old, showed that recurrent cough early in life resolved in most children.41 These children with recurrent cough without wheeze did not have AHR or atopy, and significantly differed from those with classic asthma with or without cough.41

The only randomised controlled trial on inhaled salbutamol and beclomethasone in children with cough alone showed that these medications were no different from placebo in reducing cough frequency measured objectively with a cough meter.15 The study also showed that AHR could not predict the child’s response to these asthma medications.15

Finally, the use of the term asthma needs to be interpreted in light of current controversy of wheezing phenotypes in childhood.42

Proposed model

Given the evidence outlined above, a model relating cough, CRS, and AHR in children is proposed. In a child with asthma, AHR may or may not be present.43 A trigger leads to an inflammatory response of the airway characteristic of asthma, which is clinically manifested by wheeze and dyspnoea. In some children, the cough pathway is also stimulated and these children have increased CRS during an exacerbation of their acute asthma.29

Some children without previously diagnosed asthma may present with cough. If no evidence of airway obstruction is present, most children with cough alone do not have asthma but have increased CRS.36 A small group of children have AHR but these children are no more likely to respond to inhaled salbutamol or corticosteroids than children without AHR.15 In the human airway, cough and bronchoconstriction pathways exist and both or either can be stimulated alone or simultaneously if sufficient and appropriate challenge agents are given. In children with recurrent cough who do not have evidence of airway obstruction, while both pathways are stimulated by the same trigger (such as viral infection), one pathway may be more dominant. It is not known if asymptomatic children with AHR are more likely to have hypersensitive cough receptors. A large prospective community study is required to answer this question.

Conclusions

The appropriateness of the inclusion of the symptom of cough alone as a marker of asthma severity should be questioned. Laboratory and clinical evidence indicate that cough and bronchoconstriction pathways are distinctly different in the human airway and that indices of cough (cough frequency, CRS) have no relation to airway calibre.

There is no doubt that children with asthma can present with cough, as wheeze can be underrecognised. However, recent epidemiological and clinical studies have shown that cough alone is a poor marker of asthma and indeed most children with cough do not have asthma.3 15 40 A short trial of asthma treatment may be indicated in some children, but when these are used it is important to stop treatment if it is not beneficial and not to increase the dose. If the cough is related to asthma, earlier uncontrolled trials that used theophylline suggest that the cough should relent within a week.10 12 There is no objective scientific evidence to use high or prolonged doses of inhaled corticosteroids in children with cough alone. Children with cough alone are more likely to have a temporal increase in their CRS rather than asthma.