Extraesophageal Reflux in Pediatric Patients With Upper Respiratory Symptoms

Objective  To systematically review published literature describing the association between reflux and upper airway symptoms in children.

Design  Structured MEDLINE search of English-language articles published since 1966.

Subjects  We selected articles examining reflux in conjunction with stridor, apnea, sudden infant death syndrome, life-threatening events, and laryngomalacia. Studies that focused on lower airway symptoms or adults were excluded.

Outcome Measures  Articles were abstracted for patient factors, elements of study design, methods of reflux diagnosis, and definition of pathologic reflux.

Results  Ninety-nine articles were identified, 56 of which specifically examined reflux and upper respiratory symptoms in children. Of these, 10 compared reflux incidence in symptomatic patients and a set of predetermined control patients, while the remainder reported prevalence data only. Overall, symptomatic patients were diagnosed with reflux frequently, with a range from 27% to 100%. In studies that attempted to compare patients with controls, only 2 provided statistical comparisons of the patient groups, and none adjusted for confounding owing to study design. There was marked heterogeneity in methods used to diagnose reflux in the studies reviewed, with only 34% using dual-channel pH testing; definitions of pathologic reflux were also variable.

Conclusions  Evidence seems to support the hypothesis that reflux is associated with upper airway symptoms in children. However, the strength of this correlation and the risk of upper airway symptoms attributable to reflux are difficult to determine given the limitations of available literature. Future research studies should seek standard reflux testing methods, clear comparison groups, and more rigorous statistical methods.

CONSULTATION FOR upper respiratory symptoms in children is common for the pediatric otolaryngologist. In the past, gastroesophageal reflux has been considered closely associated with stridor and other upper respiratory symptoms in children. However, to our knowledge, the evidence behind this assumption and the impact of treatment on symptoms has not been previously summarized. Although some symptoms are relatively benign and can be followed expectantly (ie, stridor from laryngomalacia without failure to thrive or significant respiratory distress), other upper airway symptoms such as respiratory distress secondary to subglottic stenosis preventing extubation in a newborn or an acute life-threatening respiratory event require more immediate investigation.

Reflux is considered a risk factor for many upper respiratory symptoms in children.¹ Additionally, otitis media, chronic sinusitis, lymphoid hyperplasia, hoarseness, laryngeal edema, or nodules have all been associated with reflux.^2-14 Therefore, identifying and treating coexistent reflux could improve outcomes while eliminating the need for invasive procedures in this population of children.

Laryngopharyngeal reflux is a more recently identified clinical entity. In contrast to gastroesophageal reflux, laryngopharyngeal reflux involves gastric acid reflux through the upper esophageal sphincter into the pharynx rather than across the lower esophageal sphincter.^15,16 As a result, laryngopharyngeal reflux can only be diagnosed by proximal (pharyngeal) pH probe placement, a technique refined in adults in the mid-1980s, but which is still not a commonly performed diagnostic modality for children despite its proposed role as the "gold standard."^15,17,18 Any episode of laryngopharyngeal reflux in adults is believed to be pathologic,^19-22 and few if any episodes of laryngopharyngeal reflux should be considered normal in children.^23-25

Laryngopharyngeal reflux has been proposed to play a key role—at least as significant as gastroesophageal reflux—in several pediatric upper airway symptoms including subglottic stenosis and acute life-threatening respiratory events. However, studies linking reflux and upper respiratory symptoms in children are heterogeneous, representing a wide range of research designs, sample sizes, definitions of reflux, and treatment protocols. Thus, the purpose of this study was to systematically review the literature with the aim of determining the strength of the available data supporting the link between reflux and upper respiratory symptoms in children.

We searched MEDLINE, the electronic bibliography database of the US National Library of Medicine, using the Medical Subject Heading (MeSH) terms gastroesophageal reflux, aspiration, esophagitis;peptic in combination with key title and text words including stridor, apnea, ALTE (acute life-threatening respiratory event), SIDS (sudden infant death syndrome), and laryngomalacia. To identify additional studies, we examined reference lists from published guidelines and book chapters in our topic area, as well as reference lists of our initially identified articles.^1,26

We then excluded non–English-language articles, those not including an abstract, those that focused solely on lower airway symptoms (such as asthma), and those that studied adults only. Articles that met these criteria were then abstracted by trained reviewers (K.W.R. and A.D.A.) for key outcomes and elements of study design. Because of the recognized difficulties in quality scoring of trials, we did not score studies meeting our inclusion criteria. However, abstraction forms for each article included key elements pertaining to trial design, such as reporting of important confounders or biases, comparability of the patient groups, and methods for matching patient groups or accounting for bias. Because 24-hour dual-channel pH probe monitoring is considered the gold standard diagnostic test for reflux, we abstracted all articles for whether this test was part of outcome measurements.^27-30 Finally, we also specifically abstracted articles for their definition of reflux.

Or analytic plan sought to distinguish articles that could be used to determine prevalence from articles that could allow estimation of risk for symptoms attributable to reflux. To this end, we split articles into 2 groups: (1) articles that reported prevalence of reflux within a defined subset of symptomatic patients and (2) articles that compared incidence of reflux in a symptomatic population with that in a predefined control group or those that attempted to match—via study design or statistical methods—reflux incidence in symptomatic and asymptomatic patients. Although all articles were reviewed for diagnostic modalities, this smaller subset of articles was then abstracted for elements that would allow estimation of the strength of the association between reflux and upper respiratory symptoms in children.

We initially identified 99 articles, 43 of which were excluded because they studied animals, enrolled adults, or focused on lower airway symptoms, yielding a core list of 56 articles that examined reflux and upper respiratory symptoms in children. We then excluded 2 studies that enrolled asymptomatic patients and reported population normative data only^31,32 and 3 case series that did not report explicit entry criteria.^33-35 Five additional studies were excluded because they involved treatment trials but did not include a control group from which conclusions regarding prevalence could be drawn.^36-40 Finally, we excluded studies in which reflux testing was performed so inconsistently that we could not discern which patients were tested and which were not,¹³ that reported results that were not clearly peer reviewed,⁵ that enrolled patients with reflux only,^12,41 that did not report objective diagnostic data,⁴² and that correlated periods of apnea in patients with known reflux.⁴³ Studies in which it was unclear whether symptoms and reflux incidences overlapped were also excluded.^44-46

Thus, our final group of articles reported results from patients enrolled in 35 studies published since 1979. As a group, these studies were uniformly based at academic medical centers or children's hospitals. Of these studies, 21 (60%) were performed prospectively and 14 (40%) were performed retrospectively, most often via medical record review. Of identified studies, only 12 (34%) explicitly reported results from dual-channel pH probe testing to diagnose reflux and 21 (60%) used other pH probe tests. Of additional tests, the most common modality was laryngoscopy and bronchoscopy (n = 13 [37%]), followed by barium swallow (n = 10 [29%]), gastric-emptying studies (n = 6 [17%]), and esophageal biopsy (n = 6 [17%]).

Of the 35 studies, 9 (26%) included subjects with multiple upper airway symptoms, 13 (37%) included subjects with apnea or a history of an acute life-threatening respiratory event, and 4 (11%) included subjects with laryngomalacia. Three studies (8%) examined patients with nasal obstruction or patients presenting with stridor from croup or subglottic stenosis. Sinusitis was the presenting symptom in 2 studies (6%), and hoarseness was the presenting symptom in 1 study (3%). Twenty-five articles reported data regarding prevalence only, and 10 attempted to compare reflux incidence in symptomatic patients with that of a control group.

Twenty-five articles reported the prevalence of reflux in patients with upper airway symptoms (Table 1). These articles displayed a range of definitions of reflux, with most defining pathologic reflux as various combinations of duration of esophageal acidity and severity of acidity. However, these definitions were inconsistent and ranged from rigorous, previously validated guidelines for diagnosis of reflux ^47,52 to those with less apparent justification.^55,65

The most common symptom examined was apnea, which was a focus of investigation in 8 (32%) of these 25 identified studies, 4 (50%) of which involved premature infants. These 4 articles reported a range between 33% and 100% in reflux prevalence in premature patients with apnea compared with a range between 25% and 100% in full-term infants with reflux. A higher reflux prevalence was noted in studies that provided results of proximal pH probe testing^29,66; lower prevalence of reflux was noted in studies that performed pH probe testing of shorter duration or that reported distal results only. A similarly broad range in reflux prevalence was noted for other upper airway symptoms, with most articles reporting prevalence of reflux in excess of 50%.

Ten articles reported prevalence of reflux in patients with upper airway symptoms and attempted to compare these patients with a control patient group (Table 2). Control groups were generally defined as asymptomatic patients or patients with symptoms not related to the upper respiratory tract. Few studies attempted any matching based on age, comorbidities (ie, prematurity), or other clinical factors, and only 2 studies ^74,78 reported any baseline information describing these factors in their patient groups. In addition, few studies reported any specific rationale for case finding or choice of controls. Similar to the studies reporting prevalence only, these articles used variable methods to diagnose reflux, with only 3 using dual-channel pH probe testing for even a proportion of patients.

Prevalence of reflux in symptomatic groups in these articles was similar to those seen in prevalence studies, with a range between 27% and 100%. Overall, most articles suggested an excess of reflux in symptomatic patients, with 7 (70%) of 10 using statistical methods to compare symptomatic patients with controls. However, other than the articles by See et al⁷⁸ and Newman et al,⁷⁴ none of these articles reported more than age and sex information regarding their patients, and none attempted to account for biases in their study design using statistical methods.

The results of our systematic review suggest that, although it is highly likely that reflux is associated with upper airway symptoms in children, the strength of this correlation is very difficult to determine. The available literature comprises studies largely derived from referral centers, most of which have substantial methodologic flaws related to inconsistent methods used for diagnosing reflux, biased patient populations, and limited statistical methods.

There is general agreement that the 24-hour dual-channel pH probe with 1 channel in the esophagus and 1 channel in the pharynx is the diagnostic study of choice for reflux.^18,79,80 Although reproducibility of pediatric 24-hour dual-channel pH probe results has been questioned⁸¹ and false-negative rates up to 20% have been reported, few other testing modalities have similar discriminative power. Barium swallow, scintiscan, endoscopy with biopsy, and bronchial washings for lipid-laden macrophages have been used,^82,83 but they demonstrate sensitivities and specificities lower than dual-channel pH probe.^84,85

Investigators have tried alternate diagnostic testing to aid diagnosis of reflux. Laryngoscopy and bronchoscopy findings—especially arytenoid edema, postglottic edema, and lingual tonsillar hypertrophy—have been demonstrated to correlate with acid exposure.^45,54 However, few data exist to correlate laryngoscopy and bronchoscopy findings with pH probe results. Esophageal biopsy is promising, but also has not been compared with dual-channel pH probe testing, nor has its safety and feasibility compared with alternate testing modalities been proven.^64,68 Also, studies using esophageal biopsy did not provide any information about pharyngeal acid exposure or laryngopharyngeal reflux.

Most studies we reviewed did not use dual-channel pH probe testing for diagnosis of reflux, and most used single-probe tests. Many studies combined results from multiple reflux tests, considering a patient to have reflux if he or she had just 1 positive diagnostic test result. Such "pooled positivity criterion" further muddies the diagnostic waters in that sequential testing may magnify the imprecision of individual tests. That is, if one orders a panel of 20 tests, each of which is 95% accurate, one can expect to have at least 1 false-positive test result and an inflated prevalence estimate. In contrast, many of the tests used for reflux have sensitivities far lower than 95% for detection of pathologic reflux, raising the possibility that these studies underestimate true prevalence.

Even within the studies that used pH probe testing, we observed a moderate amount of variability in the diagnostic criteria for reflux in terms of duration and severity of acid exposure. Although individual studies generally adhered to a published definition, across studies several criteria were used, limiting our ability to directly compare results. Even when appropriate diagnostic testing is performed, not all positive results necessarily mandate treatment. Some amount of reflux, especially in premature or newborn infants, may be physiologic and not pathologic. Comparison of different age groups would then reflect the natural history of improvement with age rather than true prevalence rates of pathologic reflux.

Many consider treatment for reflux a necessary first step in the management of various airway symptoms. In an uncontrolled trial, patients with choanal atresia who received reflux therapy formed less granulation tissue and needed fewer revision procedures.³ Another investigator found that aggressive empiric treatment of reflux in patients with subglottic stenosis led to symptomatic improvement and eliminated the need for surgical intervention.⁹ However, other studies showing no impact of reflux treatment on airway surgery for subglottic stenosis have refuted this finding.⁸⁶

Results from our systematic review suggest a reason for these apparently conflicting findings from clinical trials, since it is likely that an uncertain link between reflux and specific upper airway symptoms leads to uncertain clinical response. Empiric reflux treatment, especially when a low-risk approach is chosen, is likely to provide more benefit than risk as a first step. However, clinicians should be aware that little high-quality evidence exists to direct them to groups of patients who are most likely to benefit.

The studies we reviewed, although suggesting higher incidence of reflux in patients with upper airway symptoms, do not yet provide ample evidence to determine the magnitude of a causal link. Such a determination will require a study design using an appropriate control group, perhaps chosen on the basis of comorbidities or stage of development rather than site of care (ie, pediatric intensive care unit or referral for reflux tests on the basis of an alternate diagnosis). Admittedly, the identification and selection of controls may be difficult, but in the absence of an ideal group, adherence to standards of reporting that clinicians can use to make decisions about treatment should be the minimal standard. For example, many of the studies in our review were based at tertiary care centers and were likely subject to substantial referral bias. Few studies provided any patient information other than age or sex, making it nearly impossible to discern whether results from these studies could be used by non–tertiary care-based clinicians.^87-89 In addition to being potentially helpful in multivariable models by producing adjusted results, detailed information about patients' comorbidities would be useful for clincians simply seeking to apply evidence to their practices.

Our systematic review of studies examining the association between reflux and upper airway symptoms reveals that the literature has substantial shortcomings in terms of study design, reporting of results, methods to account for bias, and standard definitions of reflux. Although reflux plays an important role in some children with upper airway symptoms, the magnitude of this risk remains unclear. Future studies addressing these shortcomings will be required to provide the evidence for clinicians faced with these often challenging patients.

Corresponding author and reprints: Kristina W. Rosbe, MD, Department of Otolaryngology–Head & Neck Surgery, University of California, San Francisco, 400 Parnassus Ave, Suite A730, San Francisco, CA 94143-0342 (e-mail: krosbe@itsa.ucsf.edu).

Submitted for publication August 26, 2002; final revision received February 12, 2003; accepted March 12, 2003.

This study was presented at the American Society of Pediatric Otolaryngology; May 5, 2003; Nashville, Tenn.