Introduction

Chronic cough, defined as persistent coughing for more than 8 weeks, is a common clinical problem that results in significant impairment of quality of life [1]. Estimates of the prevalence of chronic cough vary widely and are as high as 33% in some epidemiologic surveys [2–5]. Although such studies may inadvertently capture some cough related to upper respiratory tract infections, associations have been identified between cough and cigarette smoking, asthma, and gastroesophageal reflux disease (GERD). The burden on health services has been studied less often, but some authors have found that 10% to 38% of those attending respiratory outpatient clinics complained of chronic cough [6, 7]. However, it must be borne in mind that many patients with chronic cough also present to otolaryngologists, allergists, and gastroenterologists.

Chronic coughing can be a feature of a multitude of conditions, and traditional dogma has required the investigating physician to determine the underlying cause by following complex diagnostic algorithms [8–10]. In the absence of an obvious diagnosis on initial assessment, such algorithms advise treatment trials and investigations targeting cough variant asthma, GERD, and nasal disease as the three most likely causes of chronic cough. A diagnosis is made when the patient’s cough responds to a specific treatment trial. Although previous publications from specialist cough clinics have suggested that such an approach is rewarded with a high level of treatment success [11], this is often not the experience of clinicians managing these patients [12]. Interest is therefore growing to help us better understand the mechanisms underlying chronic cough and how such associated diagnoses may be relevant to patients presenting primarily with severe coughing.

Chronic coughing typically affects women more than men, usually in middle age, with many patients undergoing multiple medical consultations, investigations, and trials of treatment without relief of coughing [13, 14]. Patients attending our own specialist clinic have suffered from a dry cough for an average of 5 years and generally complain of coughing in response to a range of environmental exposures (eg, smoke, temperature changes, perfumes) and to laryngeal movements (eg, laughing, shouting, prolonged talking) [15, 16]. Coughing is mediated via a persistent sensation of irritation—most commonly located in the neck/throat (75%)—associated with an “urge to cough” [15]. On direct questioning, symptoms suggesting asthma, nasal disease, and reflux disease may be elicited, but these are far less prominent than the coughing.

This review focuses on the role of gastroesophageal reflux in patients with chronic cough. The experimental and clinical evidence is summarized, and mechanisms linking reflux to cough, and possible diagnostic approaches are discussed.

What is the Evidence Linking Cough to Esophageal Reflux?

As cough and GERD are both very common, it is possible that they could coexist in patients by chance alone. A variety of approaches have been used to understand the effect of gastroesophageal reflux on both the sensitivity of the cough reflex and the amount of coughing. These are summarized below.

Effect of Infusing Acid into the Esophagus

Several studies have investigated the effects of infusing acid into the lower esophagus on cough, usually in a double-blind, randomized fashion, and comparing with a control saline infusion (Table 1). Initial studies focused on the number of coughs that occurred during the infusions, but the findings were contradictory, with one study suggesting no effect [17] and the other a significant increase in cough frequency [18]. In our experience, coughing tends to be very variable from hour to hour when measured using objective monitoring. Therefore, it is surprising that it was possible to detect a change in cough frequency over such a short period of time without very large numbers of study participants.

Table 1 Summary of cough and esophageal acid infusion studies
Full size table

Other investigators have measured the sensitivity of the cough reflex, showing that the reflex becomes significantly more sensitive in response to esophageal acid infusion [19, 20••]. The sensitivity of the cough reflex is assessed by inhaling aerosols of tussive agents, such as capsaicin or citric acid [21]. Individuals perform single inhalations of doubling concentrations of these substances from a nebulizer delivering an accurate volume of aerosolized agent when triggered by an inspiratory breath. The test is complete when the individual coughs at least five times in the 15 s after inhalation of a particular concentration of the irritant, with this concentration being referred to as the C5. Many investigators also document the concentration provoking at least two or even three coughs, known as the C2 or C3. Only C2 and C5 have been shown to be reproducible over time [22] and to correlate with other measures of cough, such as objective cough frequency [23–25] and subjective reports of cough severity [26].

Javorkova et al. [20••] and Wu et al. [19] showed that esophageal acid infusion significantly sensitized the cough reflex—as measured using the C2 and C3—in patients with chronic cough and asthma. Neither study observed spontaneous coughing during the infusions. Of particular interest, Javorkova et al. [20••] showed that esophageal acid infusion only increased cough reflex sensitivity in cough patients with coexistent esophagitis/excess esophageal acid exposure, and not those with GERD alone, suggesting that esophageal disease alone is insufficient to cause this effect. This is supported by the observation that asthma patients not suffering from chronic cough or GERD develop sensitization of their cough reflex in response to intraesophageal acid infusion [19], indicating that some airway disease may be a necessary factor. Unfortunately, studies investigating the effect of esophageal acid in other respiratory conditions are lacking.

Healthy volunteers in these esophageal acid infusion studies consistently failed to cough or develop sensitization of their cough reflex. This contrasts with pain studies in which esophageal acid infusion has been shown to sensitize both visceral and somatic pathways [27].

Measurement of Esophageal Reflux in Chronic Cough

Esophageal pH Studies

Many studies quantifying reflux in patients with chronic cough select participants in whom other common causes of cough (eg, asthma, nasal disease) have been excluded, thus assuming that reflux is the most likely cause in these individuals. Early studies in limited numbers of cough patients referred for esophageal studies suggested they have slightly increased esophageal acid exposure compared with age-matched controls [28]. Furthermore, those reporting a response to GERD treatment tended to have greater esophageal acid exposure than nonresponders [29].

Recent larger studies, however, have found pH monitoring to be of limited clinical value. Among 146 patients attending a specialist cough clinic, 56.2% had an abnormal test (excessive acid exposure/positive symptom index [SI]), and of those receiving reflux therapy, only 40.8% responded [30]. Moreover, telephone follow-up after 2 years revealed only 28.5% had maintained their initial response, and no parameters of pH monitoring accurately predicted this response. In a similar study of 53 chronic cough patients referred for pH monitoring, 54.7% had an abnormal result (excessive esophageal acid exposure/positive symptom association probability [SAP]/positive SI), of which only 35.8% reported relief following treatment [31]. Treatment response was significantly predicted by female gender, positive SAP, and the intensity of reflux therapy, but not acid exposure time or SI (see section on Temporal Association Studies [Methodologic Issues] for definitions of SAP and SI). Unfortunately, the sensitivity, specificity, and positive predictive value of the SAP were poor (47%, 82%, and 28%, respectively). There are some difficulties in interpreting these observations when data are limited on patient selection, and those with good treatment responses may have been more likely to respond to requests for information several years later.

Esophageal Impedance Studies

The development of esophageal impedance/pH monitoring has allowed more detailed assessment of reflux in terms of detecting non-acid (pH > 4) as well as acid reflux events (pH < 4), proximal extent, and nature (ie, liquid/gaseous/mixed). Studies of patients in whom extraesophageal causes of cough were excluded found that the average numbers of acid/non-acid reflux events (liquid ± gas) detected were not outside the 95th percentile for healthy volunteers, with few reaching the proximal esophagus [32–34]. Similarly, in unselected chronic cough patients, we found that most had numbers of reflux events within the normal limits despite the average number of events being slightly higher than in healthy controls [35]. In addition, the proportion of acid, non-acid, and proximally occurring esophageal events did not differ between patients and controls.

Pharyngeal Impedance Studies

Two studies have assessed reflux extending into the pharynx using impedance/pH measurements in patients with chronic cough. One study compared nonasthmatic chronic cough patients with asthma patients and found only a median of two liquid (±gas) episodes (interquartile range, 1–4) per 24 h per individual in the pharynx [36]. All episodes apart from two had a pH greater than four, and there was no difference between the diagnostic groups. In a second more recent study, gaseous esophagopharyngeal reflux was compared among chronic cough patients who reported that their cough responded to proton pump inhibitor (PPI) therapy (now not on PPIs), patients with typical GERD symptoms only, and healthy volunteers [37•]. Although again there were no differences in the numbers of non-acid gaseous pharyngeal reflux events (pH change <1 unit) between groups, patients with cough did exhibit slightly more weakly acidic gaseous reflux, (median, 2 per 24 h [pH fall >1 unit with nadir >4]). These events appeared to associate with belching and swallowing, and in our experience, both commonly occur as a consequence of sipping fluids, a technique often used to suppress chronic cough, and thus may be an epiphenomenon. Why these events were weakly acidic is difficult to explain, other than patients responding to acid suppression were selected.

Temporal Association Studies

Methodologic Issues

In addition to documenting the number of reflux events, their duration, and proximal extent, many studies already discussed also assessed whether reflux events were temporally associated with episodes of coughing, using various mathematical approaches to analyze these relationships (eg, the SI and SAP). The SI is defined as the proportion of cough events that occurred within 2 min of a reflux event, and in which a value of 50% or greater represents a positive association [38]. The SAP calculates whether cough events follow reflux events more frequently than would be expected by chance alone using the Fisher exact test and the standard 2-min window, with an SAP of 95% or greater assumed to indicate a positive association [39]. However, some caution must be exercised when using these measures, and their limitations must be borne in mind, as discussed in detail by Connor and Richter [40].

The timing of coughing episodes is usually recorded by instructing patients to press an event marker each time they cough, as is traditionally the case when associating reflux symptoms with reflux events. However, in contrast to symptoms such as heartburn, several hundred coughs typically occur over a 24-h period in patients with chronic cough [14], making the task of remembering to mark each event very difficult. Even if individuals are advised to only mark each burst of coughing, there can still be a large number, which might lead to underreporting. For example, we have shown using a device that objectively records cough that these patients exhibit coughing bursts a median of 120 times (interquartile range, 69.6–170.4) per 24 h [41].

This contrasts with studies describing the number of coughing episodes marked by button press, which have reported an average of 20 to 30 coughing episodes per 24 h [32, 33], with as few as 0 to 3 events per 24 h in some studies [42]. When direct comparisons between button press and objective measures of cough have been made, not only have most coughing episodes not been marked, but a delay also has been noted before marking occurred [33, 43]. A delay will tend to underestimate both reflux–cough associations (as 2 min may have elapsed before marking occurs) and cough–reflux associations (cough preceding reflux may not be marked until after the reflux event). Finally, one additional issue that has recently come to light is the observation that the presence of a nasogastric catheter reduces coughing by approximately one third [42]. The reason for this is unknown but unfortunately cannot be avoided if temporal associations are to be made, not only with acid, but also with non-acid reflux events. This section therefore focuses on studies in which objective assessment of the precise timing of cough events has been performed (summarized in Table 2).

Table 2 Temporal associations between reflux and cough using objective measurementsa
Full size table

Reflux–Cough Associations

All recent studies investigating the temporal associations between cough and reflux (assessed by the SAP using a 2-min window [32, 33, 42, 44••]) have suggested that in a significant proportion of chronic cough patients (20%–48%), cough follows reflux more frequently than expected by chance alone. Moreover, the pH of the reflux events preceding cough does not seem to matter, with events of pH greater than four equally important as those with pH less than four [44••, 45•].

Most of these studies have relied on ambulatory esophageal manometry for the detection of coughing. This technique has some limitations in that most investigators only feel confident to identify epochs of coughing (ie, with several pressure swings) rather than single cough episodes, and its accuracy to identify events has not been validated. We have developed an ambulatory system for recording cough sounds over 24 h [46–48] and in a recent study have reported a greater prevalence of reflux–cough/cough–reflux associations in patients with chronic cough than had been found previously [44••]. For the purposes of such analysis, we quantified coughing in cough epochs, defined as a period of continuous coughing without a 2-s pause. Using this technique combined with simultaneous esophageal pH/impedance, we recorded a mean of 59.2 cough epochs and 67.7 reflux events per 24-h period (the reduced number of cough epochs compared with our previous study [37•] was due to the presence of nasogastric catheter [49] and exclusion of coughs during mealtimes). It is interesting that the increased accuracy of cough identification led to a greater detection of temporal associations, similar to another recent study using a different cough sound–recording device overnight [45•]. Although it could be argued that increasing the number of coughs detected simply increased the probability of chance associations, with an average of only two to three cough epochs and two to three reflux episodes occurring per hour, there is plenty of time for these events to fall outside the 2-min window.

Whether a 2-min time window is appropriate for studying reflux–cough associations is a matter of debate and requires further study to determine the best window that describes the physiologic time required to translate reflux to cough, which could be affected by the pH of the refluxate [50]. The original rationale for the 2-min time window was based on the time taken to press an event marker after experiencing chest pain [51]; therefore, it may not be optimal in studies using objective cough measurements. However, that several investigators have found associations over a 2-min period lends some validity to this time frame, and the standardization has facilitated comparisons between studies.

Cough–Reflux Associations

Most studies have also identified patients with the reverse association of cough preceding reflux events. Again we detected more cough–reflux associations using our ambulatory cough sound–recording device than previously identified using manometry, although Kunsch and colleagues [45•], using technology similar to ours, found no significant cough–reflux associations overnight. Whether sleep or the supine posture influence this relationship is not known.

In addition to using the 2-min window, we also investigated the effect of shortening the time window to just 10 sec for cough–reflux associations to determine whether the straining that occurs during a coughing episode provokes reflux. Use of a 10-sec window resulted in far fewer patients exhibiting cough–reflux associations, suggesting that straining is an unlikely cause of cough-induced reflux and leading us to speculate that coughing instead may be inducing transient lower esophageal sphincter relaxations and, consequently, reflux.

Endoscopic Findings

Despite a limited number of studies, endoscopy of the esophagus generally has been assumed to have a low yield in patients presenting with chronic cough, and is not recommended in diagnostic guidelines. This is perhaps not surprising when most patients with typical symptoms of reflux disease have no evidence of erosive disease at endoscopy. In our study of unselected patients with chronic cough, we found that only 11 patients (18%) had erosive esophagitis (grade I, n = 10; grade II, n = 1), with this being no more prevalent in those with a positive compared with a negative SAP [44••]. In contrast, however, in a study of 405 patients with suspected GERD presenting with otolaryngology symptoms, 52.3% were found to have erosive disease, compared with 38.4% of control patients with typical reflux symptoms. Those presenting with “predominant cough” had the highest prevalence of erosive disease (65.7%) [52]. Given the latter observation, it is difficult to completely rule out a role for erosive esophagitis in patients with chronic cough, and it is important to note that patients in that study had not undergone a trial of acid suppression therapy. We find that many patients referred to our specialist cough clinic have already tried PPI therapy, and because it may take time for the esophagitis to recur [53], it could be that the prevalence of esophagitis in our study was artificially reduced. Regardless, the identification of erosive disease in otolaryngology patients was not useful in predicting treatment responses [54].

Very few data exist with respect to microscopic changes in the esophageal mucosa among patients with chronic cough. In patients with nonerosive reflux disease, although biopsy findings are nonspecific, dilated intracellular spaces (DIS) are a frequent finding, as are increased numbers of inflammatory cells, basal cell hyperplasia, and papillary elongation of rete pegs [55]. It has been speculated recently that DIS may play a role in chronic cough [56]. However, why these patients would experience coughing rather than heartburn as a result of DIS is difficult to explain, especially when many patients with chronic cough present with “silent reflux” (ie, without heartburn or other typical symptoms).

Esophageal Motility Disorders

Several studies have suggested an increased prevalence of esophageal motility disorders in patients with chronic cough [57–59], most commonly ineffective esophageal motility. To date, these findings have not been found to be of any clinical utility, and while ineffective motility could lead to more prolonged exposure of the esophagus to reflux, this has not been investigated.

What Pathophysiologic Mechanisms Link Cough and Reflux?

There are three potential sites in which refluxate could evoke coughing: 1) the larynx, 2) the bronchi, and 3) within the esophagus. At each site, reflux theoretically could directly trigger cough or sensitize the cough reflex, increasing the likelihood of coughing being provoked by usually innocuous stimuli.

Airway

Laryngopharyngeal Reflux

Refluxate reaching the larynx and pharynx is thought to underlie a variety of symptoms, including globus, hoarseness of the voice, throat clearing, and chronic cough, and is frequently referred to as laryngopharyngeal reflux (LPR). It has been suggested that these symptoms result from injury and inflammation caused by the acid and pepsin in the refluxate contacting the laryngeal tissues. Clinically, LPR is often diagnosed by examination of the larynx for particular clinical signs of inflammation, but these signs show poor interobserver agreement and could be present in patients with chronic cough as a consequence of the trauma of coughing [60]. Because no diagnostic test has been identified that confirms the presence of LPR or predicts a response to reflux therapy, diagnosing LPR in chronic cough is challenging.

Although evidence has shown that refluxate occasionally reaches the proximal esophagus [32, 35] and more rarely the pharynx (average, 2 per 24 h) in chronic cough patients [36], it does so no more frequently than what occurs in healthy controls [61]. Although it is possible that the pharynx may be more sensitive to these “normal” amounts of reflux, it is difficult to imagine that such infrequent events (ie, ~2–3 per 24 h) could be responsible for severe coughing.

Microaspiration

Although microaspiration of refluxate into the airways often has been suggested as a potential mechanism for cough, there are very few data to support or refute this hypothesis. For microaspiration to occur, refluxate must first travel up into the pharynx/larynx and second breach the airway’s protective mechanisms, including coughing. In those with neurological deficits, disordered swallowing also may be a factor. Moreover, it is important to recognize that “pharyngeal reflux” is not synonymous with “microaspiration” and has been detected even in the pharynx of healthy individuals [61]. However, as previously described, very few reflux events reach the pharynx in chronic cough patients, and unless airway protection is impaired, these events are unlikely to be aspirated.

With respect to airway protection, some studies have suggested that the larynx may be a less-than-perfect barrier, as 50% of healthy volunteers were found to microaspirate a radio-labeled nasopharyngeal infusion during sleep [62], although these test conditions are far from normal circumstances. Because cough is one of the main protective laryngeal reflexes, it would seem counterintuitive that patients with chronic coughing would microaspirate more than healthy individuals, and we recently showed that cough frequency is inversely related to sputum pepsin concentration [35]. Furthermore, it has been extensively documented that these patients are more sensitive to the inhalation of irritant chemicals during cough reflex sensitivity testing, and report coughing in response to minor exposure to irritants, suggesting heightened airway protection. We also have shown that airway pepsin concentrations in bronchoalveolar lavage samples taken from chronic cough patients are no different from those in healthy volunteers [35]. More compelling evidence for microaspiration playing a role in respiratory disease comes from lung transplant recipients [63–65], who are known to have a diminished cough reflex following surgery [66].

The only studies to suggest impaired airway protection in patients with chronic cough and associated GERD are those showing a diminished laryngeal adductor reflex compared with that of healthy controls [67, 68]. Although an absent laryngeal adductor reflex is associated with a risk of aspiration in patients with dysphagia [69], whether reduced sensitivity conveys a risk of microaspiration in chronic cough patients who have normal swallowing and a heightened cough reflex is unknown. In addition, the lack of control patients with GERD (without cough) in these studies makes it difficult to determine whether the diminished laryngeal adductor reflex relates to the reflux disease or to the cough.

Esophagus

Neuronal Sensitization

As discussed previously, gastroesophageal reflux plays a role in coughing in up to half of the individuals with chronic cough. The sensitizing effect of esophageal acid infusion on the cough reflex and the fact that gastric refluxate into the distal esophagus temporally associates with cough both support a neuronal mechanism linking reflux to cough. Furthermore, that those patients exhibiting reflux–cough associations do not have an increased number of reflux events or excessive esophageal acid exposure compared with normal limits [33, 44••] suggests that this mechanism, once established, may continue to be triggered by physiologic levels of reflux.

The shared vagal innervation of the cough reflex and esophagus provides the most likely pathway through which esophageal stimuli may lead to coughing (Fig. 1). This is often referred to as an esophagobronchial reflex. However, whether such a mechanism exists under normal physiologic conditions and thus constitutes a “real reflex” remains unclear. For example, in healthy volunteers, whereas rapid esophageal distention triggers an esophagoglottal closure reflex, it does not cause coughing [70]. In addition, infusing acid into the esophagi of healthy volunteers fails to evoke coughing or to sensitize the cough reflex [18, 19, 20••]. Perhaps a better analogy would be cross-organ sensitization, as described between the organs within the pelvis. Inflammatory processes in the colon/rectum, for example, have been shown to induce increased pain sensitivity in the bladder, and vice versa [71]. The convergence of afferent fibers in spinal nerves and on second-order neurons in the spinal cord is thought to be the basis for which inflammation in one organ can sensitize responses from another [72]. These mechanisms resemble those described in the viscerosomatic referral of pain, and mediating central sensitization in pain hypersensitivity states.

Fig. 1
figure 1

Schematic showing the stimuli known to trigger coughing in patients with chronic cough and afferent vagal pathways. nTS—nucleus tractus solitarius

Full size image

Thus, it is conceivable that in patients with chronic cough, airway disease may cross-sensitize the esophagus, and similarly, esophageal disease may cross-sensitize the airways. However, unlike the cross-sensitization between the pelvic organs, in which convergence of spinal nerves is responsible for transmitting painful stimuli, only interactions between afferents in the vagus or in the nucleus tractus solitarius of the brainstem could explain communication between the esophagus and the cough reflex. Our finding that cough reflex sensitivity was heightened in chronic cough patients with a positive SAP (compared with those with a negative SAP) [44••] is in keeping with this hypothesis of cross-organ sensitization, although it does not provide any indication of the nature of the sensitizing insult or whether this originated in the esophagus or the airway. Moreover, many chronic cough patients also display cough–reflux associations that do not appear to be caused by cough-induced straining, suggesting a further role for cross-organ sensitization. Perhaps vagal cross-organ sensitization allows coughing to provoke transient lower esophageal sphincter relaxation and thus a cough–reflux event, although this has yet to be studied.

Which Diagnostic Approaches?

There is little evidence to suggest that any objective measures of reflux in patients with chronic cough can predict a response to therapy; therefore, it is difficult to recommend investigation of such patients.

Despite a relatively high yield of abnormal pH monitoring in patients with chronic cough, acid exposure time does not seem to be a useful predictor of response to acid suppression therapy [30, 31]. SAP is the only parameter to show any relationship to the success of therapy, but only in a retrospective study, in which although it was statistically significant, the relationship was too poorly predictive of responses for use in the management of individual patients [31]. One small study of six patients has suggested SI may be predictive of a response to fundoplication, but the study was uncontrolled with no systematic assessment of coughing [73].

Furthermore, although several studies have identified patients with objective evidence of temporal associations suggesting that reflux causes cough, these measurements currently rely on non–commercially available cough-recording equipment. Moreover, as no placebo-controlled trials of acid suppressants have been conducted in patients with positive SAP, is it not yet clear whether such a therapeutic approach is helpful, especially when the numbers of reflux events in these patients are often within normal limits and non-acid events are just as likely to be linked to coughing as acid events. Perhaps interventions that target neuronal sensitization or reduce the number of reflux events may be a more effective approach. Randomized controlled trials using objective measures of cough in patients selected for evidence of reflux causing cough (ie, SAP positive) are lacking [74]. Therefore, until further evidence becomes available, the current approach of treatment trials of acid suppression therapy, as advised by cough management guidelines, must be recommended.

Conclusions

In conclusion, most patients with chronic cough have no more gastroesophageal reflux than healthy volunteers in terms of numbers of events; acidity; or extension of refluxate into the proximal esophagus, larynx, or pharynx. Moreover, there is little indication of microaspiration. However, observations that 1) cough temporally associates with distal reflux events; 2) that in these individuals, cough reflex sensitivity is heightened; and 3) esophageal acid infusion increases cough reflex sensitivity in patients with cough and reflux disease, but not in those with reflux alone, suggest that a central neuronal mechanism links cough to reflux. This may explain in part the poor efficacy of traditional drugs, such as those suppressing gastric acidity. Thus, new targets need to be identified and appropriate agents developed.