Results

A total of 134 consecutive patients with asthma were invited to enter a Veterans Affairs study on the treatment of gastro-oesophageal reflux. All patients were offered upper gastrointestinal endoscopy and oesophageal function tests. Six patients (4.5%) declined to participate. The pH tracings of the remaining 128 patients were analysed. During the pH metry, patients did not experience more symptoms than usual, and the 24 hour recording periods proved to be typical of their presenting symptomatology. Episodes of coughing or wheezing occurred in 60 and 25 of 128 patients, respectively. The number of coughing episodes ranged between 1 and 79 (mean (SD) 21 (3)), and the number of wheezing episodes ranged between 1 and 27 (8 (1)). Among patients who coughed, five had only one cough and two had only two coughs during the entire monitoring period. Among patients with wheezing, six had one or two wheezing episodes only. These 13 patients were excluded from the subsequent statistical analyses. No abnormal oesophageal motility was recorded in all but one patient with low amplitude contractions. (This patient had only three coughs and no wheezes during the 24 hour pH metry and was not included in the further analysis.) A flow diagram of all patients is shown in fig 1. Table 1contains the characteristics of the entire patient population and the two subgroups of 53 patients with frequent coughs and 19 patients with frequent wheezes. Demographic characteristics, consumption of alcohol and cigarettes, acid contact time, prevalence of oesophagitis, and hiatal hernia were similar in all patient groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow diagram of the study population.

Table 2 shows the temporal relationships between coughing or wheezing and acid reflux. Assuming a latency period of two minutes or less, acid reflux preceded a coughing episode about seven times more often than vice versa—that is, coughing preceded acid reflux. Both types of events occurred about 2–3 times more frequently than expected based on chance alone, the outcome being statistically significant. Overall, nearly half of all coughs were found to be associated with a reflux episode. An increase in the latency period from two to five minutes increased the percentage of associations, with reflux preceding cough remaining the far more frequent relationship. About half of all wheezing episodes were associated with acid reflux (table 2). In 43% of all wheezing episodes, reflux preceded wheezing. Because of the relatively small number of patients with frequent wheezing, we abstained from a more detailed statistical analysis of the temporal relationship between wheezing and reflux.

Of all reflux, coughing, and wheezing events, 81%, 78% and 80%, respectively, were recorded during the upright position. In general, separate analysis of the pH metry data by upright versus supine body position revealed rather similar patterns of association. Assuming a latency period of two minutes, 49% versus 36% of all upright versus supine coughs were associated with reflux events, reflux preceding cough in 43% and 30%, respectively. Of all wheezes, 51% and 41% were associated with reflux events during the upright and supine periods, respectively. Assuming a latency period of five minutes, 60% versus 71% of all upright versus supine coughs were associated with reflux events, reflux preceding cough in 48% and 58%, respectively. Of all wheezes, 57% and 53% were associated with reflux events during the upright and supine periods, respectively.

Figure 2 contains a sensitivity analysis of varying the length of interaction between cough and reflux from one to 10 minutes. Progressive lengthening of the coughing period raises the chances for a reflux event to fall into that period. As the duration of influence exerted by coughing is artificially extended from one to 10 minutes, an increasing fraction of coughs appears to result in a reflux episode (fig 2A). Assuming the influence of coughing on reflux to last as long as 10 minutes after the coughing event was first recorded, about 15% of all coughs appear to be followed by a reflux episode. In fig 2B, a reflux episode is assumed to exert its influence on coughing 1–10 minutes after the pH value returned to pH >4. The fraction of reflux episodes leading to coughing increases accordingly from 14% to 25%. Most importantly, the sensitivity analysis shows that the measured strength of the relationship between coughing and reflux depends in part on the type of assumptions built into the analysis. The analysis also demonstrates that, regardless of such assumptions, more reflux episodes result in coughing than coughing episodes result in reflux. When each individual patient is considered, a similar pattern is revealed. Table 3 shows the association of coughing and reflux in each individual patient. In 49 of 53 patients, acid reflux preceded a coughing episode while in 28 of 53 patients a cough preceded a reflux episode. All patients with cough preceding reflux also had reflux preceding cough. In only four patients was there no relationship between the two events.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Fraction of coughs preceding reflux (A) and fraction of reflux episodes preceding coughs (B) dependent on the length of the latency period of coughing and reflux episodes, respectively. Each point represents the mean (SD) value calculated from the average values of 53 individual patients.

In the baseline analysis, a minimum of five coughing episodes was chosen as the threshold value for including a patient in the statistical analysis. In the sensitivity analysis of fig 3, the threshold was raised progressively from five to 35 coughs per 24 hour recording period. The chosen threshold had no influence on the overall fraction of coughs leading to acid reflux (fig 3A). In patients with rare as well as frequent coughing, only a minority of coughs resulted in acid reflux. The fraction of reflux episodes leading to coughing increased as the threshold was raised from five to 35 coughs per 24 hour recording period (fig 3B). Such a pattern suggests that in patients with frequent coughs the likelihood of coughs being precipitated by acid reflux increases. Varying the threshold of total wheezes between three and 10 did not greatly affect the fraction of wheezes associated with reflux and was similar to the baseline value of 48%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Influence of the number of coughing episodes on the fraction of coughs preceding reflux (A) and fraction of reflux episodes preceding coughs (B). Only patients with coughing episodes equal to or exceeding the threshold shown on the horizontal axes were included in calculating the mean (SD) value.

Discussion

Gastro-oesophageal reflux has long been associated with pulmonary disease.25-27 By analysing 24 hour pH tracings of asthmatic patients, we showed that half of all episodes of coughing or wheezing were associated with acid reflux. Acid reflux was far more likely to precede than follow a coughing episode. This temporal sequence between reflux and cough suggested that the reflux event was frequently responsible for coughing and that only rarely did coughing cause acid reflux. The likelihood of reflux induced coughing increased as the number of coughs increased in the individual asthmatic patient.

The study population comprised an unselected series of adult asthmatics in a VA hospital who were referred from the outpatient medicine and pulmonary clinics to participate in a study on the treatment of gastro-oesophageal reflux in asthmatics. Patients represented the entire spectrum with respect to pulmonary disease severity. A bias towards more or less severe asthma was unlikely as almost all patients approached agreed to participate in the study. Twenty eight per cent of all patients did not use bronchodilators on a permanent basis, indicating an appreciable fraction of patients with relatively mild or intermittent disease only. Placebo controlled studies using prolonged pH monitoring have failed to show any effect of either theophylline or β agonists on the severity of gastro-oesophageal reflux.28 ,29 Our patients therefore were instructed to use their bronchodilators as needed. Cessation of therapy might have resulted in bronchoconstriction, hyperinflation, flattening of the diaphragm, and a subsequent increase in acid reflux.30Indeed, an increase in gastro-oesophageal reflux has been demonstrated during methacholine induced bronchospasm.17 Thus any potential beneficial effect on acid reflux gained by discontinuation of bronchodilator medications could be negated by the adverse effects of worsening dyspnoea on acid reflux.

The primary aim of our study was to evaluate the temporal relationship between acid reflux and cough and make inferences about their causal association. Patients underwent extensive training on how to use the event markers for coughing and wheezing episodes. During evaluation of the pH metry data it became obvious nevertheless that even a seemingly simple analysis of the temporal relationships was open to multiple types of bias. With frequent consecutive and overlapping episodes of reflux and cough, it was difficult to establish unequivocally a cause and effect pattern. Two minutes were chosen as a baseline latency period between two events because this time period had been reported previously to provide reliable estimates.21 ,22 If two events occurred within a two minute time frame, the first event was considered to have caused the second one. Because other investigators also used latency periods as long as five and 10 minutes,23 ,24 we decided to subject our data to a computer analysis that automatically assigned to all latency periods lengths that varied between one and 10 minutes. Although the fraction of observed associations increased with increasing latency periods, the overall pattern remained robust. Regardless of the length of the latency period, reflux of acid remained a more common event to precede cough than to follow a cough. For clinical purposes therefore the choice of a single latency period of two minutes would probably be sufficient to establish a relationship between pulmonary symptoms and reflux.21

In a second sensitivity analysis, thresholds for the total number of coughs and wheezes were varied. Patients with very few events were characterised by extreme R values and extreme fractions of coughs or wheezes associated with reflux. The thresholds of three and five events for wheezes and coughs, respectively, were chosen as a compromise between two opposing goals. On the one hand, a large number of events assured a more reliable estimate for the true fraction of associated events in the individual patient. On the other hand, the very stringent criteria of high thresholds led to exclusion of many patients. Fortunately, most parameters remained unaffected by the choice of a particular threshold. Thus the dependence of reflux induced coughing on the total number of coughs appeared to reflect an underlying pathophysiological mechanism rather than an artefact of measurement.

Fifty three of our 60 patients with cough had five or more coughing episodes during the 24 hour period, and 92% of the 53 patients had cough associated with reflux. These findings are consistent with the results of DeMeester et al and Ing et al who reported that 70% and 90%, respectively, of all patients with chronic cough also presented with abnormal pH monitoring.14 ,31 Laukkaet al attempted to establish a causal relationship between coughing and reflux in a small population of 10 patients with gastro-oesophageal reflux symptoms.32Although in the majority of instances the two events were not related, in a few instances cough preceded reflux more often than reflux preceded cough. Compared with our study however the authors investigated only a few patients, most of whom had a low acid contact time with little evidence to support reflux induced pulmonary symptoms. Raiha et al observed in a series of 105 patients undergoing oesophageal pH monitoring that respiratory symptoms did not appear unless the acid contact time of pH <4.0 exceeded 10%.33 It is presently unknown whether reflux of acid into the oesophagus is sufficient to initiate a cough or whether actual aspiration of the refluxed acid is necessary. Indeed, changes in airway calibre and resistance in response to acid in the distal oesophagus have been described in asthmatic patients as a possible reflex mechanism for reflux induced cough.34 ,35 Our results support recently published data by Harding et al who showed that among asthmatics with reflux disease, 78% of respiratory symptoms and 90% of coughs were associated with oesophageal exposure to acid.36 The statistical associations between pulmonary symptoms and acid reflux found by us and by previous investigators do not prove causality and still require confirmation by treatment trials. Studies demonstrating improvement or even cure of asthma after antireflux surgery however do lend support to the contention that gastro-oesophageal reflux contributes to the occurrence of asthma.8 ,10

In conclusion, half of all coughs and wheezes in asthmatics are associated with acid reflux into the oesophagus. While an occasional coughing episode can lead to reflux, it is the reflux episode, in the vast majority of instances, that leads to cough. Our study establishes a possible link between reflux and coughing or wheezing. The potential of antireflux therapy to break the cycle of reflux induced cough and, to a lesser degree, cough induced reflux is in need of study.