Objective There are no objective ambulatory studies on the temporal relationship between reflux and cough in children. Commercial pHmetry loggers have slow capture rates (0.25 Hz) that limit objective quantification of reflux and cough. The authors aimed to evaluate if there is a temporal association between cough and acid pH in ambulatory children with chronic cough.
Design, setting and patients The authors studied children (aged <14 years) with chronic cough, suspected of acid reflux and considered for pHmetry using a specifically built ambulatory pHmetry–cough logger that enabled the simultaneous ambulatory recording of cough and pH with a fast (10 Hz) capture rate.
Main outcome measures Coughs within (before and after) 10, 30, 60 and 120 s of a reflux episode (pH<4 for >0.5 s).
Results Analysis of 5628 coughs in 20 children. Most coughs (83.9%) were independent of a reflux event. Cough–reflux (median 19, IQR 3–45) and reflux–cough (24.5, 13–51) sequences were equally likely to occur within 120 s. Within the 10 and 30 s time frame, reflux–cough (10 s=median 2.5, IQR 0–7.25; 30 s=6.5, 1.25–22.25) sequences were significantly less frequent than reflux–no cough (10 s=27, IQR 15–65; 30 s=24.5, 14.5–55.5) sequences, (p=0.0001 and p=0.001, respectively). No differences were found for 60 and 120 s time frame. Cough–reflux sequence (median 1.0, IQR 0–8) within 10 s was significantly less (p=0.0001) than no cough–reflux sequences (median 29.5, 15–67), within 30 s (p=0.006) and 60 s (p=0.048) but not within 120 s (p=0.47).
Conclusions In children with chronic cough and suspected of having gastro-oesophageal reflux disease, the temporal relationship between acid reflux and cough is unlikely causal.
Statistics from Altmetric.com
As chronic cough causes considerable morbidity, it is not surprising that cough is a common reason for parents to seek medical help,1 and acid-suppressing agents are commonly used. Acid-related gastro-oesophageal reflux disease (GERD) is one of the most common causes of chronic cough in adults.2 Indeed, one study reported that all adult subjects with GERD coughed during infusion of acid.3 However, other studies suggest that GER is associated with, but not the causative factor for, cough.4 There is disparity of opinion in the causality link between cough and GER.5 Although adult respiratory-based cough guidelines advocate a trial of empirical treatment of GERD even in the absence of GER symptoms,2 gastroenterology-based6,–,8 and paediatric cough guidelines9 10 are less definitive about the association between cough and GERD. Thus, although the common coexistence of cough and acid GERD is established, objective studies ascertaining cause and effect are limited. Reasons for this include the difficulty of defining the occurrence by chance of two common symptoms, the changing definition of GERD, lack of randomised controlled trials and equipment limitations.5 11
What is already known on this topic
▶ Instillation of acid into the oesophagus may provoke cough, but cough can also provoke a reflux event. Thus, acid-suppressing agents are often empirically used for chronic cough particularly in adults.
▶ Cough and acid reflux commonly coexist, and causality is often assumed.
What this study adds
▶ From the analysis of 5628 coughs recorded using a novel pH–cough logger, we found that most coughs occurred independently of a reflux event and cough–reflux and reflux–cough events were statistically equally likely to occur.
▶ In children with chronic cough and suspected of having GERD, the temporal relationship between acid reflux and cough is unlikely causal.
Most commercial pHmetry systems have a maximum capture or download rate of 0.25 Hz (ie, data points recorded once every 4 s). The active respiratory muscle phase of a single cough epoch lasts 0.6 to 0.8 s, and the glottic closure phase of cough whereby the greatest intrathoracic pressure is generated lasts 0.2 s.12 Thus, to objectively ascertain if cough occurs before a reflux event, an instrument with a capture rate of at least 5 Hz (one data point every 0.2 s) is a prerequisite. Furthermore, data captured on synchronised (eg, to the nearest second) separate instruments as opposed to a single time frame will give erroneous results given that resolution rates have to be less than the compressive phase of a cough when intrathoracic pressures peak up to 300 mm Hg,12 as it is the phase most likely associated with a reflux event.
Studies showing the association between cough and GER13 have been primarily based on the subjective cough reporting, which is less reliable than objective data.14 15 Few ambulatory studies exist that evaluate the relationship between cough and reflux objectively. Indeed, there are no published ambulatory objective studies in children, and the only objective study in children showed that cough induces reflux.16 All ambulatory objective studies to date have used manometry17,–,20 to quantify cough that is based on alterations in intraoesophageal pressure reflecting the increase in intrathoracic pressures that occur with the expiratory phase of cough. However, similar intrathoracic pressures can occur with other events such as sneezes. Although there is some disagreement over this between pulmonologists and gastroenterologists,21 objective cough assessments do require ambulatory validation,15 which have not been performed for oesophageal manometry-defined coughs.
As there is no suitable commercial equipment, we specifically developed an ambulatory pHmetry–cough logger that uniquely enabled the simultaneous ambulatory recording of cough epochs and pH with a fast capture, recording rate and response time.5 Using this instrument, we studied prospectively children with chronic cough considered for 24 h pHmetry for GER assessment, to evaluate if a temporal association of cough with acid pH occurs in children. We hypothesised that in children with cough, coughs are more likely to precede, than follow, acid reflux events (pH falls to <4).
Children aged 3 months to 14 years presenting to paediatric gastroenterologists at our institution were recruited when they were considered for an ambulatory 24 h pHmetry to investigate suspected GERD. Ten of the children were referred to the gastroenterologists by respiratory doctors for assessment of GER. All children had a current chronic cough.22 Children on any type of acid suppression in the preceding 7 days before pHmetry, previous fundoplication or gastro-oesophageal surgery, acute asthma or any known underlying chronic lung disease (other than episodic asthma) were excluded. The study was approved by the human ethics committees of the Royal Children Hospital and University of Queensland. Informed consent was obtained from one parent. In children aged >12 years, an additional child-specific consent was obtained.
Ambulatory pHmetry–cough logger
A solid-state ambulatory cough meter (consists of three electromyographic (EMG) electrodes and a microphone) that we have developed and described previously23 was augmented with ambulatory 24 h pHmetry. This enabled simultaneous measurement of pH and cough signals. In addition to concurrent pHmetry, the cough meter component of our pHmetry–cough logger differed from our previous version23 by the following features: (1) use of non-volatile memory (a multimedia card) for primary storage of data captured, (2) digitally captured and stored data, (3) adjustable sensitivity thus enabling its use in young children who have lower amplitude EMG and cough audio signals and (4) capture rate of 10 Hz. Thus, the functionality of the two channels of the cough meter previously designed and validated23 was improved. Figure 1A,B depicts our pH-metry–cough logger. The calibration and physical plug enabled use of a standard commercial oesophageal pH probe (Zinetics 24 h paediatric pH probe; Medtronic). The pH probe was placed in the standard paediatric manner, 3 to 5 cm above the lower oesophageal sphincter using Stroebbels formula and radiologically verified.
The interface from the pHmetry–cough logger to the computer was via a multimedia card. The downloaded data were converted and plotted onto a chart program (Maclab Chart; ADInstruments, Castle Hill, New South Wales, Australia), which produced images previously published.23 Coughs and pH changes were recorded by a researcher blinded to the child's medical data. The cough-meter component was previously validated.23 We checked the pHmetry component against standard clinical equipment used in our hospital (Medtronic Functional Diagnostics, Copenhagen, Denmark) and verified that measurements of pH using several buffers for 8 h were accurate (data not shown). Before use, the pHmetry component was calibrated using buffer pH 4 and 7 solutions.
A reflux episode was defined as a fall in pH to <4 lasting ≥0.5 s. The possible scenarios were recorded: cough before reflux episode (cough–reflux sequence; fig 2), cough after reflux episode (reflux–cough sequence), reflux episode with no preceding cough (no cough–reflux sequence), reflux episode with no cough after reflux episode (reflux–no cough sequence) and isolated cough (cough with no reflux episode 120 s before or after cough). All possible scenarios were recorded within 10, 30, 60 and 120 s before and after a reflux episode. A maximum of 120 s was chosen based on the upper quartile cough latency period (mean 70 s, approximate interquartile range (IQR) 20–120) in a study where all 22 adults with GER had cough with hydrochloric acid infusion.3 Other studies have also utilised this time frame.17 18 The sequences were scored consecutively starting from commencement of the study. Thus, if the first event was a pH fall, sequences around that event was scored, and the next event was considered an independent event for scoring purposes so as to not bias the results.
As data did not have a normal distribution, non-parametric methods were used for all analysis, Kruskal–Wallis and Wilcoxon tests for paired and unpaired data and Pearson χ2 test for categorical variables. Post hoc analysis was performed to examine an age effect. All descriptive data are expressed as median and IQR. Two-tailed p<0.05 was considered significant. SPSS v.13 was utilised for all statistical calculation. Given the absence of any data in children, we could not calculate a sample size a priori but planned to enrol a similar number of subjects to previous adult studies17 19 that examined the temporal relationship between cough and reflux episodes.
Twenty-two children were recruited, but in two children, recordings were unsuccessful (pH probe dislodged in one (child vomited) and the EMG tracings inadequate from dislodgement in one). The median age of the 20 children with complete data was 1.1 years (IQR 0.66–4.2), 10 boys and 10 girls. The indication for pHmetry in all children was suspicion of reflux, and symptoms present in the children were history of regurgitation (n=13), difficulty in swallowing (n=10), halitosis (n=6) and vomiting (n=12). The children had a total of 5628 coughs that were analysed. The median number of coughs was 159 (IQR 57–402) per child and that of pH falls was 31.5 (IQR 15.25–68.75). The frequencies of cough–reflux, reflux–cough, no cough–reflux and reflux–no cough sequences are shown in table 1. The respiratory history, severity of cough as scored on the scale and previous use of medications are presented into table 2. Given the small number of children, we did not perform any regression analysis on these factors.
Reflux and cough associations
Overall, 83.9% of coughs were independent of a reflux episode within 120 s. The median number of coughs that was not associated with any reflux episode (within 120 s) was 140 (IQR 48.75–324.75). The mean and median percentages of cough associated with any reflux episode (cough–reflux or reflux–cough sequences) were 16.1% and 4.6% (IQR 6%–22.3%) respectively.
The frequencies of cough–reflux to reflux–cough sequences within 10, 30, 60 and 120 s of the reflux event were similar (p range 0.23–0.87; table 1). In addition, there was no significant difference in the number of no cough–reflux to reflux–no cough sequences for all the time frames (p range 0.34–0.55; table 1). Expressed as a percentage of total coughs, reflux–cough (13.27%) were similar to cough–reflux (11.38%) sequences at the 120 s time frame.
Evaluating reflux as the primary event for cough
Reflux–cough sequences (2.5, IQR 0–7.25) within 10 s were significantly less (p=0.0001) than reflux–no cough sequences (median 27, IQR 15–65) and within 30 s (p=0.011) but not within 60 or 120 s (p=0.07 and p=0.42 respectively; table 1). Only 13.27% of coughs occurred 120 s after reflux. The median ratios of reflux–cough to reflux–no cough sequences were 0.037 (IQR 0–0.28) for 10 s, 0.235 (IQR 0.026–0.643) for 30 s, 0.247 (IQR 0.026–1) for 60 s and 0.605 (IQR 0.093–1.642) for 120 s.
Evaluating cough as the primary event for reflux
Cough–reflux sequence (median 1.0, IQR 0–8) within 10 s was significantly less (p=0.0001) than no cough–reflux sequences (median 29.5, IQR 15–67), within 30 s (p=0.006) and 60 s (0.048) but not within 120 s (p=0.47; table 1). The median numbers of coughs not followed by a reflux event (cough–no reflux sequence) within 10, 30, 60 and 120 s were 158.5, 157.5, 151.5 and 145, respectively; expressed as a percentage of total coughs, the respective numbers are 99.6%, 98.1%, 95.6% and 92.9%.
Subanalysis of data
Data were subanalysed post hoc based on age (infants aged ≤12 months);compared with the rest of the cohort (10 in each group), we found no differences between total number of coughs, number of coughs not within 120 s of a pH fall or any one of the sequences described previously (p range 0.27–0.94, data not shown).
We used a novel ambulatory pHmetry–cough logger that enabled the accurate categorisation whether cough preceded, or followed, a pH fall to <4 within 10, 30, 60 and 120 s. From the analysis of 5628 coughs in 20 children with chronic cough and suspected of having GERD, we found that the mean percentage of cough associated with any reflux episode (cough–reflux or reflux–cough sequences) was only 16.1%. That is, most cough occurred independent of a reflux event. Cough–reflux and reflux–cough were equally likely to occur. Indeed, within 10 and 30 s, reflux–cough sequences were significantly less frequent than reflux–no cough sequences. We found also that a reflux episode (cough–reflux) was less likely than no reflux episode after a cough within 10, 30 and 60 s.
Relating pHmetry to subjective reporting of cough has been established for several decades.24 However, there are few objective studies and no ambulatory studies in children. The sole objective study in children used scintigraphy16 and examined non-spontaneous cough (ie, cough on command). Our current study is the first in children in an ambulatory real-life setting that has quantified the relationship between acid GER episodes and cough objectively. In our cohort, children were equally likely to have a cough–reflux or reflux–cough sequence. Thus, our data suggest that when quantified objectively, the temporal relationship between cough and acid reflux occurs by chance (as opposed to temporally causative). Although our study is small, it is very similar to other prospective studies17 19 that studied the objective relationship between cough and reflux intensively. The median number of coughs in our cohort of 159 was much higher than that reported by some (median 3,18 mean 30,25 mean 2417) but in the similar range to others (mean 12620). We thus analysed many more cough episodes in our total cohort (5628 coughs) compared with those in other studies (reported number of total coughs analysed in whole cohort of other studies were 64717 and 49818). A type II error in any study is always possible. However, this as the cause of the negative findings in our study is unlikely given the large number of cough and reflux episodes analysed and the similar numbers to other studies.17 19 Moreover, we found that cough after a reflux episode (reflux–cough sequence) was less frequent than reflux–no cough sequences within 30 s and equally likely to occur in the 60 and 120 s time frame.
Paterson and Murat19 were the first to relate cough and pH changes in ambulatory adults objectively. Using manometry (as opposed to a validated cough meter) in 15 adults with chronic cough, they reported that 13% of coughs were induced by reflux and 1.6% of reflux events precipitated cough.19 Sifrim et al17 also used manometry in 22 adults and reported that 69.4% of coughs were independent of reflux and 30.6% occurred within 2 min of a reflux episode. In a retrospective study of 51 adults, Bogte et al18 found that 29.5% of coughs were acid reflux related, of which 28.6% were reflux–cough sequence and 71.4% were reflux–cough sequence. In the study by Blondeau et al,20 the largest prospective report of adults with chronic cough, acid GER was identified as the potential cause in half of the 100 patients evaluated. However, all these studies were in adults, and it is now established that chronic cough in children differs from that in adults.22 26 In addition, none of these studies examined statistically whether the proportion of reflux–cough sequence were similar to that of reflux–no cough episodes. In our study, we found that the number of cough–reflux sequence was similar to that of reflux–cough sequence, which suggests a chance occurrence in young children with chronic cough. When two events are common, the chance occurrence of coexistence is high. Cough has been described to be present in up to 81% of children with biopsy-proven oesophagitis.27
Our study also differed from the adult studies in the definition of a reflux event. There is no consistent period (for pH<4) used for defining a reflux event. In studies that have objectively examined the temporal relationship, the period used varied from >4,17 20 >5,19 25 to >10 s.18 We elected to choose a short period (>0.5 s) based on the studies that have elicited immediate cough on infusion of intraoesophageal acid.3 28
We did not study laryngopharyngeal reflux, which is widely regarded by some (particularly ear, nose and throat (ENT) specialists) as a cause of chronic cough related to GERD. However to date, all but one randomised controlled trial with subjects enrolled from ENT clinics where cough was an outcome measure have shown that proton pump inhibitor (PPI) therapy (the mainstay of GERD treatment) is not better than placebo.11 29 This includes the largest (n=145) of these studies involving GERD therapy with cough as an outcome measure.30 Two independent systematic reviews found that high-dose PPI therapy was no more effective than placebo in producing symptomatic improvement or resolution of cough presumed related to laryngopharyngeal reflux.29 31 Furthermore, a controlled non-randomised study showed that fundoplication was not efficacious for cough either.32 In contrast, “uncontrolled studies suggest that 40–100% of patients who have suspected acid-related ENT symptoms improve on aggressive anti-reflux therapy.”33
Reflux of gastric contents into the oesophagus can be acidic, weakly acidic17 or weakly alkaline (non-acid reflux) and includes volume reflux. We have only examined acid reflux given that PPIs are the most commonly used therapy for GERD. Thus, our study is limited to acid reflux and cannot be extrapolated to non-acid reflux, which arguably requires oesophageal impedance studies. In addition, it is unknown if acute effects related to cough preceding or following a reflux event are relevant to the management of chronic cough. The pathophysiology relating cough to GER is likely multifactorial and includes upregulation of cough neuropathway,34 neurogenic inflammation,35 direct acid stimulation of vagal receptors present in the oesophagus28 and aspiration of the refluxate.
Although the common coexistence of cough and GERD is well established, our study raises further questions about causality between acid reflux and cough in children, as also articulated in the latest American and European GER guidelines for children.8 PPIs are the current standard therapy for acid reflux and have a highly acceptable safety profile. However, there is increasing recognition of their adverse effects (in children and adults)36,–,39 and has cost implications. Thus, further studies to delineate who will benefit from acid suppression are important to guide best practice. The sample size in our study was far too small to perform a valid regression analysis of the five possible sequences against phenotypic features of the children or to compare age groups. Ideally, a randomised controlled trial should be doneto determine if patients with predominant cough–reflux sequence respond differently to those with reflux–cough sequence. This is particularly pertinent in infants where the efficacy of 4 weeks of PPIs for symptoms attributed to GERD is not only doubtful but also increases respiratory adverse events.38 In children with chronic cough and suspected of having GERD, the temporal relationship between acid reflux and cough is unlikely causal.
The authors thank the parents and children who participated in this study. The authors are also grateful to Mary Da Silva and the nurses of the Department of Gastroenterology, Royal Children's Hospital. This study was funded by the Royal Children's Hospital Foundation. ABC is supported by the Royal Children's Hospital Foundation and by a practitioner fellowship from the National Health and Medical Research Council (grant number 525216) and a Queensland Smart State Clinical Fellowship.
Funding Royal Children's Hospital Foundation.
Competing interests None.
Ethics approval This study was conducted with the approval of the Royal Children's Hospital and University of Queensland.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.