Temporal Association of Polysomnographic Cardiorespiratory Events With GER Detected by MII-pH Probe in the Premature Infant at Term

Nunez, Jeanne*; Cristofalo, Elizabeth*; McGinley, Brian; Katz, Richard; Glen, Daniel R§; Gauda, Estelle*


Figure 2 reproduced incorrectly in the May 2011 issue of the journal. The correct presentation is as follows:

Journal of Pediatric Gastroenterology and Nutrition. 52(6):787, June 2011.

Journal of Pediatric Gastroenterology & Nutrition: May 2011 - Volume 52 - Issue 5 - p 523–531
doi: 10.1097/MPG.0b013e3181fa06d7
Original Articles: Gastroenterology

Objectives: The aim of the study was to examine temporal association (TA) between polysomnographic cardiorespiratory (CR) events and gastroesophageal reflux (GER) in premature infants with persistent CR events at >39 weeks postmenstrual age and determine whether the use of multichannel intraluminal impedance (MII)-pH probe improves sensitivity of the TA compared with pH probe alone.

Patients and Methods: Seven infants born between 24 and 29 weeks' gestational age with persistent CR events at 39 to 48 weeks' postmenstrual age underwent a polysomnography with MII-pH probe. Symptom index (SI) and symptom-associated probability were calculated for diverse types of reflux and CR events. SI and a Fisher exact test with variable association windows were calculated for obstructive apnea (OA). Odds ratios for an OA given a reflux event and for a reflux event given an OA were determined.

Results: With a Fisher exact test, a subject-specific association between MII events and OA was found in the 3 patients who required a fundoplication or had the worse clinical GER. Some level of TA was found with SI and symptom-associated probability in 6 of 7 infants. Association was found for pH >4 and pH ≤4 reflux events. pH-only events with no change of MII had only a limited role in generating CR events.

Conclusions: TA between CR events and GER was found in a single-subject–level analysis in some infants with persistent CR events at term. This TA suggests a causal relation between CR and reflux events that was further strengthened by the clinical outcomes of each infant.

*Department of Pediatrics, Neonatology, Johns Hopkins Hospital and School of Medicine, Baltimore, USA

Department of Pediatric Pulmonology and Sleep Medicine, Johns Hopkins Hospital, Baltimore, USA

Division of Pediatric Gastroenterology, Johns Hopkins University, the Mount Washington Pediatric Hospital, Baltimore, USA

§Scientific and Statistical Computing Core of the NIMH Intramural Research Program, Bethesda, Maryland, USA.

Received 2 September, 2009

Accepted 28 August, 2010

Address correspondence and reprints requests to Jeanne Nunez, MD, Johns Hopkins Bayview Medical Center Neonatal Intensive Care Nursery, AA2 Building, 4940 Eastern Ave, Baltimore, MD 21224-2780 (e-mail: jnunez4@jhmi.edu).

This work was supported by a Thomas Wilson grant.

Daniel Glen is affiliated with the National Institute of Mental Health, National Institute of Health, but has contributed to this article in a private capacity. The authors report no conflicts of interest. The views expressed in the article do not necessarily represent the views of NIH or the United States.

Article Outline

Cardiorespiratory (CR) events persist frequently beyond 39 weeks' postmenstrual age (PMA) in premature infants (1,2) and are associated with various morbidities that include acute life-threatening events, apneas, feeding dysfunction, and possibly worse developmental outcome (2–4). Gastroesophageal reflux (GER) is often considered the cause of CR events. This association has led to interventions including a gastrostomy with fundoplication when GER medications fail to stop the CR events. At a neonatal intensive care unit step-down unit in Baltimore, Maryland, 41% of infants born at <25-weeks' gestation received surgically placed gastric feeding tubes with or without fundoplication (5). Fundoplications are commonly performed in infants (6) suspected clinically of having GER with reported subsequent improvement of chronic lung disease (CLD) and persistent CR events (7), therefore suggesting a cause-and-effect relation between GER and CR events. Indications for fundoplication are generally based on clinical history (persistent CR events beyond term) and abnormal pH probe studies when other causes of CR events have been ruled out because demonstrating a direct relation between apnea and GER has been difficult. Studies attempting to temporally associate CR events and GER have been conflicting, secondary to major experimental differences that have included methods of investigation, patient selection, and temporal association (TA) statistical analyses (8–17). Additionally, esophageal pH monitoring, the standard method of diagnosing GER in adults, will not detect nonacid reflux, the most common type of reflux seen in infants (18). Multichannel intraluminal impedance (MII)-pH probe monitoring allows the tracking of a fluid bolus inside the esophagus giving the bolus direction, velocity, and pH level. The MII method is based on the variability of the electrical resistance of air, muscle, and fluids to an alternating current generated between multiple pairs of electrodes. The MII method is sensitive, and consistent association with manometric events has demonstrated good specificity as well (19,20).

To date, the literature has included no reports of polysomnographic (PSG) studies with concurrent MII-pH probe monitoring to evaluate premature infants with persistent clinical CR events occurring beyond term PMA. Earlier studies used pH probes only and/or focused on younger premature infants still at a significant risk of apnea of prematurity and older infants with history of acute life-threatening events or chronic pulmonary disease (8–17).

The purpose of the present study was to examine the TA between PSG events and episodes of reflux in a select group of premature newborn infants exhibiting persistent CR events at >39 weeks' PMA during presurgical evaluation for fundoplication. We tested the hypothesis that the use of MII-pH probe improves sensitivity of the TA compared with the pH probe alone, and we determined whether it is potentially useful as a clinical tool to identify the cause of persistent CR events in former premature infants.

Back to Top | Article Outline



The present study was conducted at the Johns Hopkins Hospital and Mount Washington Pediatric Hospital in Baltimore, Maryland, between May 30, 2006 and July 24, 2007. Informed consent was obtained from parents. The Johns Hopkins institutional review board approved the study protocol.

We studied 7 premature infants born between 23 and 29 weeks' gestation with persistent clinical CR events at 39 to 48 weeks' PMA. Clinical CR events were detected by the bedside CR monitor and defined as episodes of heart rate at <80/min and/or oxyhemoglobin desaturations to <90% and/or apnea lasting >20 seconds. None of the infants had major congenital anomalies, and all had normal upper gastrointestinal (UGI) studies to rule out anatomic abnormalities of the UGI tract that may cause symptoms similar to those of GER disease. All of the infants were clinically well and stable at the time of the study, and all of them were fed either orally or with a nasogastric feeding tube placed solely for bolus feedings.

Back to Top | Article Outline


Antacid reflux medications (H2 receptor antagonists, proton pump inhibitors) were stopped 5 days before the study. An overnight standard PSG was performed with a transnasal MII-pH probe. The MII-pH probe (Comfortec MII-pH probe, ZIN-BS-51, Sandhill Scientific, Highlands Ranch, CO), designed for esophageal lengths ≤15 cm, has 6 impedance channels separated by 1.5 cm with a pH sensor between the 2 lowest impedance channels. The MII-pH probe is a 6.4-F catheter. We confirmed by x-ray that the pH sensor was placed 2 cm above the lower esophageal sphincter.

The PSG and MII-pH probe analysis systems were synchronized by digitally marking each tracing at the beginning of the study. The specific times of each PSG CR event were adjusted for the offset between the 2 clocks of the PSG and MII-pH probe recordings. An additional pair of markings was repeated during the PSG to confirm the offset between the 2 clocks. Any additional shift of time during PSG was considered negligible compared with the association window size. The impedance-pH data were analyzed using the Sandhill analysis software and visually validated.

Reflux events were defined by at least a 50% decrease in impedance from baseline in at least 2 distal channels with a characteristic tracing on successive impedance channels that indicated the direction of the fluid bolus upward, from the stomach to the esophagus. Data during feeding were excluded from the analysis because of the difficulty of differentiating reflux events in the presence of frequent swallowing. The MII-pH probe method allowed us to determine the following: MII all, MII-pH ≤4, and MII-pH >4 reflux events; all pH ≤4 events—pH ≤4 events with or without changes of MII; pH-only events—pH ≤4 events with no change of MII; and MII or pH-only events. Standard PSG was performed simultaneously in all of the patients to monitor sleep state and respiration patterns. The recorded and analyzed signals included electroencephalogram, right and left electrooculogram, submental electromyography (EMG), electrocardiogram, chest and abdominal wall motion, arterial oxygen saturation by pulse oximetry, end-tidal carbon dioxide, and anterior tibialis leg EMG. Airflow signals were recorded from both nasal cannula and nasooro thermistor. All of the signals were acquired digitally (Alice 4; Respironics/Philip Andover, MA, or Somnologica/Embla, Broomfield, CO) with a technician present throughout the study. To determine whether the nasal MII-pH probe may have been obstructive, the probe was removed for the last 3 hours of the study.

CR events were recorded by the PSG software and scored according to the standards of the American Academy of Sleep Medicine (21). We used the following definitions:

* Apnea: Complete cessation of airflow for at least 2 breaths. Each respiratory event was subclassified as either central apnea (CA) or obstructive apnea (OA) based on the assessment of the respiratory flow and effort signals (abdominal and thoracic plethysmography). An event was scored as a mixed apnea (MA) when the airflow signal met duration and amplitude criteria for OA, associated both with and without respiratory effort.

* CA: Complete cessation of airflow for 20 seconds or after cessation of airflow for >2 breaths associated with either an arousal from sleep or oxyhemoglobin desaturation ≥3%.

* Hypopnea: >30% reduction of airflow associated with either an arousal from sleep or oxyhemoglobin desaturation ≥3%.

* Microarousal: Electroencephalogram shift to a faster frequency that includes theta, alpha, and beta frequencies ≥16 Hz, but not spindles after a minimum of 10 consecutive seconds of stable sleep. Arousals in rapid eye movement were scored only if accompanied by an increase in submental EMG.

Some PSG apneas may be extremely short because they may last only 2 breaths. We assumed that these shorter, discrete apneas detected by PSG were indicative of an increased risk and are the baseline of the critical CR events detected by the bedside monitor.

Back to Top | Article Outline

Statistical Analysis of Temporal Association

Association between reflux and CR events was determined using the following statistical measures:

1. Symptom index (SI) is the percentage of symptoms (in this case, CR events) related to reflux events. This test is included in the Sandhill Scientific software package (version 5.1.2). In this software package, a symptom is considered to be correlated when a reflux event occurs within a 5-minute window preceding the symptom. SI is considered significant when ≥75% and equivocal when between 50% and 75%. Rather than using a fixed window size, we examined the association between the 2 types of events with window sizes of 60, 120, 180, 240, and 300 seconds preceding the OA using the SAS statistical analysis package (SAS Institute, Cary NC).

2. Symptom-associated probability (SAP) Weusten is a 2-tailed Fisher exact test, newly modified by Sandhill Scientific according to the method proposed by Bredenoord et al (22). The null hypothesis tested with a Fisher exact test is that the CR and reflux events occur randomly with a uniform distribution, and that the probability of a reflux event preceding a CR event occurs by chance alone. The temporal data from the entire procedure are divided into 2-minute bins that are classified for the presence of reflux and apneic events. A symptom is considered to be associated when the onset of a reflux event occurs within a 2-minute window before the symptom. The “R+” is the number of unique bins that contain reflux episodes with “R−” as the complement, that is, the number of bins with no reflux episodes. Similarly, symptoms (apneic events) are categorized as S+ and S− events. The combination of symptoms following reflux events within the time association window are considered to be associated (S+R+) events. These events are considered only when the leading edge of any reflux episode intersects the symptom association window; otherwise, they are counted as (S+R−). (S−R+) is calculated as (R+) − (S+R+), the number of bins in which symptoms did not occur and reflux was present. (S−R−) is calculated as (R−) − (S+R−), the number of bins in which neither symptoms nor reflux events were present. SAP is considered statistically significant when ≥95%.

3. We also analyzed the association between OA and reflux events using a right-tailed Fisher exact test designed according to the Sandhill specifications of the SAP Weusten as described above. We used both Microsoft Excel software and an online Fisher exact test (http://www.langsrud.com/fisher.htm) to compute the TA with varying window sizes. We varied the bin size and window size accordingly from 15 to 300 seconds and calculated the Fisher exact test for MII all, MII-pH ≤4, MII-pH >4, MII or pH only, and pH-only events.

4. Finally, we analyzed the association between OA and reflux events with logistic regression and identified odds ratios to have an OA given a reflux event and a reflux event given an OA for the entire group of patients.

Back to Top | Article Outline


Description/Demographics of Patients


We studied 7 premature infants born between 23 and 29 weeks' gestation with persistent clinical CR events at 39 to 48 weeks' PMA (Table 1). Only 1 infant was still receiving oxygen; the others had been without supplemental O2 for 1 to 8 weeks before the study. All of the patients were receiving chronic diuretics and antireflux medications (either proton pump inhibitors or H2 receptor antagonists and prokinetic agents). All of the infants had received no caffeine since approximately 35 weeks' PMA. Hematocrit at the time of the study was between 30% and 39%. The mmHg level of PCO2 was in the 40s for all of the patients except for patient 5, whose PCO2 was in the 60s and who was still receiving O2 and being studied for persistent oxyhemoglobin desaturations alone without other CR events. Of the 7 infants, 4 had at least a grade II IVH. Five of the 7 infants showed reflux on UGI studies. UGI including a swallow study in patient 1 showed frequent nasopharyngeal reflux, laryngeal penetrations, and a single episode of minimal silent aspiration during swallowing, with delayed emptying of the esophagus, and a small sliding hiatal hernia. The PSG respiratory disturbance index was 18.6, 46, 23.3, 12.2, 16, 0.3, and 32 for patients 1 to 7, respectively.

Back to Top | Article Outline

Poststudy Outcomes

Our decision to perform a fundoplication was based not on the results of TA between CR events and reflux events but instead on clinical history (persistent CR events beyond 42 weeks' PMA), and pH probe study results with other causes of CR events ruled out.

Patient 1: A gastrostomy with fundoplication was performed, after which the CR events improved. At 1-year follow-up, the gastrostomy tube had fallen out, was not replaced, and the infant was growing well.

Patient 2: The infant's apneas improved spontaneously. A repeat PSG 1 month later showed improvement, but at 2 years of age, the infant continued to have severe GER. Upper endoscopy showed esophagitis. At 2 years of age, the patient's asthma was reported to be worse with GER symptom aggravation, and he was treated with proton pump inhibitors, thickeners, and H2 receptor antagonist.

Patient 3: The infant underwent a fundoplication and ventriculoperitoneal shunt revision for posthemorrhagic hydrocephalus. CR events improved after fundoplication despite multiple VP shunt malfunctions and sepsis.

Patient 4: The infant improved without fundoplication on antireflux medications. A repeat PSG 1 year later showed mild obstructive sleep apneas with an respiratory disturbance index of 3.6/hour.

Patient 5: Desaturations and hypercarbia were thought to be explained on the basis of CLD. The infant received a gastrostomy without fundoplication and was discharged home on O2.

Patient 6: The patient CR events improved without a fundoplication. Proton pump inhibitors and prokinetic agents had been discontinued by the private pediatrician. At 17-month follow-up with the pediatric pulmonologist, the infant had reactive airway disease, which according to parents remained unchanged with the discontinuation of GER medications. The parents were advised to watch for symptoms of GER disease. Reactive airway disease was attributed to CLD.

Patient 7: The patient CR events improved without a fundoplication. One month later, a repeat PSG showed similar results with severe obstructive sleep apneas associated with mild hypoxemia and desaturation on the basis of lung disease. At 2-year follow-up, the infant had no recent respiratory problems.

Back to Top | Article Outline

Polysomnography CR and Reflux Event Results

Of the 196 OA events found across all of the patients, 83% occurred in active sleep, with a mean duration of 6.811 ± 2.92 seconds. Of the 78 MAs, 94% occurred in active sleep, with a mean duration of 9.47 ± 2.99 seconds. Of the 25 CAs, 92% occurred in active sleep, with a mean duration of 6.06 ± 2.26 seconds. Of the 168 desaturations, 81% occurred in active sleep, with a mean duration of 11.06 ± 5.32 seconds. Finally, of the 362 microarousals, 60% occurred in active sleep, with a mean duration of 14.86 ± 10.17 seconds.

Of the 159 cases of pH-only events, 64% occurred in patient 1 (Tables 2 and 3). Additionally, a total of 173 episodes of MII reflux events occurred, of which 87% were liquid, 13% were mixed liquid and gaseous, 48% had a pH ≤4, and 52% had a pH >4.

Back to Top | Article Outline

Temporal Association

Table 4 provides a description of the SI for OA (with a 300-second window of association) (Sandhill software). Results of a right-tailed Fisher exact test are also given in Figures 1 to 4 for OA and reflux events. For each patient, results of SI, SAP, and Fisher exact test are provided in the text below.

Back to Top | Article Outline

Symptom Index

SI (association window of 300 seconds) values were ≥50% in patient 1 for OAs, CAs and hypopneas, desaturations, and microarousals, associated with MII all and all pH ≤4 events, and for MAs with all pH ≤4 events. In patient 2, SI was ≥50% for OA and MAs associated with MII-pH >4 only. No association was found with all pH ≤4 reflux events. In patient 3, SI was ≥ 50% for OA and hypopneas associated with MII all, all pH ≤4, MII or pH-only events, and for desaturations associated with all pH ≤4 events. In patient 6, SI was ≥50% for desaturations associated with MII all and all pH ≤4 events; and in patient 7, SI was ≥ 50% for MAs associated with MII all. SI was <50% for all CR events in patients 4 and 5.

By varying the association window size between OA and reflux events in patient 1, SI values for OA and reflux events detected by MII or pH-only events were 48.6% at 60 seconds, 70% at 120 seconds, 79% at 180 and 240 seconds, and 81% at 300 seconds. In patient 2, SI values for OA and reflux events detected by MII or pH-only events were 22% at 60 seconds, 34% at 120 seconds, 37% at 180 seconds, 43% at 240 seconds, and 50% at 300 seconds. In patient 3, SI values for OA and reflux events detected by MII or pH-only events were 12% at 60 seconds, 24% at 120 seconds, 40% at 180 seconds, 52% at 240 seconds, and 56% at 300 seconds.

Back to Top | Article Outline


SAP was significant at ≥95% for patient 1 for OA associated with MII-pH ≤4, MII-pH >4, MII all, and all pH ≤4 events; CAs associated with MII-pH ≤4; desaturations associated with MII-pH >4; and microarousals associated with MII-pH >4. For patient 2 for OAs and MAs associated with GER detected by MII all only; for patient 3 for OAs and desaturations associated with MII all events and hypopneas associated with MII-pH ≤4 and all pH ≤4 events; for patient 4 only for desaturations associated with MII-pH ≤4; and for patient 7 for OAs associated with MII-pH ≤4 events. SAP was not significant for any CR events for patient 6.

Back to Top | Article Outline

Fisher Exact Test

By varying the window size of the Fisher exact test, we found that in patient 1 there was significant TA as early as 15 seconds for MII all, MII or pH-only events, and pH-only events. The lowest P value was obtained at 120 seconds (Fig. 1). TA for MII-pH ≤4 occurred as early as 60 seconds and for MII-pH >4 at 90 seconds. The lowest P values were also obtained for MII-pH ≤4 and MII-pH >4 at around 120 seconds. The addition of the less-acidic events, MII-pH >4, improved the TA when compared with MII-pH ≤4 alone (Fig. 4). With the Fisher exact test in patient 2, TA was significant as early as 45 seconds for MII all. The lowest P value was obtained at 90 seconds (Fig. 1). There was no association noted with pH-only events with a P value of 1 (Fig. 2). Only 1 MII event had a pH ≤4; the rest of the MII events had a pH >4. In patient 3, TA was significant as early as 15 seconds for MII all. The lowest P value was obtained at 120 seconds (Fig. 1). There was no significant association for pH-only events for all of the windows of association (Fig. 2). TA was significant as early as 15 seconds for MII-pH ≤4 and 105 seconds for MII-pH >4. The addition of MII-pH >4 improved the TA when compared with MII-pH ≤4 alone (Fig. 5).

By using a Fisher exact test, TA was not significant for any window of association in patient 7 (Figs. 1 and 2) and in patients 4, 5, and 6. Patients 4 and 6 did not have any OA occurring within a 300-second association window, and patient 5 did not have any OA during the entire PSG.

The addition of pH-only events to MII weakened the TA, as measured using the Fisher exact test, for patients 2 and 3 (Fig. 2). In patient 1, the addition of pH-only events weakened the TA for the majority of windows of association (Fig. 3). In patients 1, 2, and 3, the TA between OA and MII-pH ≤4 occurred at earlier windows of association when compared with MII-pH >4 (Figs. 1, 4 and 5).

Of the 7 patients recruited for this study, 6 patients showed some significant level of TA. In contrast, we observed no TA in only 1 patient (patient 5), who was recruited for persistent desaturations with an O2 requirement and no other CR events.

Back to Top | Article Outline

Group Data Statistical Analysis

The odds ratios of having an OA given any reflux event and having a reflux event given an OA were not significant for any window of association between 60 and 300 seconds.

In summary, we have reported some level of TA between CR and reflux events using SI, SAP, and the Fisher exact test in 6 of the 7 infants studied. Of these 6 infants, the 3 infants who had the most consistently elevated SI >50% for diverse CR events and significant Fisher exact test for OA also had the worst clinical GER as demonstrated by their outcomes. Two of these infants needed a fundoplication and subsequently improved, and the third infant's CR events improved but continued to have severe clinical GER. The 3 other infants for whom we found only an occasional association with desaturations or MAs using SI and OA using SAP for MII-pH ≤4 had a good outcome at 2 years of age without a fundoplication with no history of further CR events or GER. The last infant who showed no significant TA also had a good outcome and improved without a fundoplication.

Back to Top | Article Outline


The present study is the first to our knowledge to document the TA of reflux with CR events in this specific population of former premature infants with persistent CR events at term. At this age, the risk of apnea of prematurity is decreased, and other causes of CR events including GER are considered. This is the first attempt to our knowledge to explore the TA for potential clinical use in these infants during presurgical evaluation for fundoplication. The population of former premature infants at term with persistent CR events is heterogeneous and somewhat rare. Although our study population was small, the statistical method we used demonstrates TA for each specific child in a single-subject–level analysis. Showing a TA does not necessarily demonstrate a cause-and-effect relation; however, our criteria for fundoplication, based on clinical history and pH probe alone together with the respective outcomes (ie, improvement with fundoplication or no surgical treatment), were supported by the results of the TA. This relation of outcome to both the traditional clinical indications and these TA statistical measures gives strong credibility to the usefulness of this method as a clinical diagnostic tool. Additional studies are needed to confirm the relation between TA and outcome. Additionally, in the presence of TA, it still remains to be determined whether the reflux-associated apneas will resolve spontaneously with maturation.

The SI may not be adequate to analyze TA for CR events, especially with a 5-minute association window. Furthermore, SI was designed for pain that generally has only a single cause, whereas this may not be the case for CR events. The SAP and Fisher exact tests are more appropriate methods because these tests determine the probabilities of the observed CR and reflux event pairs occurring randomly.

Our Fisher exact test is similar to the SAP Weusten test in the Sandhill software. They share the same null hypothesis and are carried out using a Fisher exact test based on a 2 × 2 matrix of binned events, but our implementation has several significant improvements. First, the Fisher exact test is right-tailed, which is more appropriate for these data. It also allowed us to obtain a P value instead of a percentage that responds to the statistical question more directly than the complement percentage of the SAP. More important, our implementation allows us to look at TA by varying both bin and association window sizes. Finally, this test allowed us to look at the TA with diverse types of reflux events detected by MII-pH probe (the Sandhill implementation of SAP does not allow a look at pH-only events).

We found no TA by using logistic regression on the entire population, a somewhat expected result given the heterogeneous nature of the subject population. In the present study, each infant had varying degrees of clinical symptoms with no control group to compare against, and so no single group statistic is likely to show significance. For this reason, the data were analyzed principally using statistical measures appropriate for single-subject analysis.

Combined MII-pH probe and PSG are the most sensitive methods available to detect reflux and CR events, respectively. The combined MII-pH probe makes it possible to detect pH ≤4 and pH >4 bolus MII reflux as well as pH-only events detected only by pH probe. By using a flow sensor at the nose, the PSG allows for characterization of the type of apneas (central, mixed, obstructive, and hypopneas) during each sleep stage. In the present study, we examined TA for the first time not only for the OA but also for MAs, CAs, desaturations, and microarousals. By using a Fisher exact test, we found an association in 3 patients for OA, also variably by using SI and SAP from 1 infant to another, with CA, MAs, hyponeas, desaturations, and microarousals. Furthermore, we were able to find a subject-specific association with pH >4 and pH ≤4 reflux events or both.

Back to Top | Article Outline

Meaning of pH-only Events

Using a Fisher exact test, we found an association between OA and reflux events detected by MII in some infants; however, in these infants, we found that the TA between OA and reflux was weakened (increased P value) by the addition of pH-only events and was not significant for pH-only events in some patients. These data suggest that pH-only events (which are most likely extremely short reflux events not reaching >2 impedance rings or clearance/swallow of previous reflux event (18,23)) may have a limited role for some patients in terms of the association with OA.

Cases of pH-only events have been observed commonly in our and other studies (9,18,23,24). In our study, 171 episodes of reflux events and 235 episodes of pH ≤4 were detected by MII only. Of these totals, 67% were pH-only and 33% were associated with changes of MII. These results may provide an explanation of why a TA has been difficult to demonstrate in previous studies using a pH probe alone and why the MII-pH probe improves sensitivity of the TA compared with the pH probe alone. Our results are consistent with another recent study (23) in which pH-only events were found to have a higher pH and more rapid clearance compared with the classic 2-phase acid reflux event detected by MII, suggesting that pH-only events are less likely to result in clinical manifestations.

Back to Top | Article Outline

Differences From Prior Studies

Peter et al (12) found no TA, but they studied younger premature infants from 24 to 34 weeks' gestation still at significant risk for apnea of prematurity. In these cases, TA may be underestimated if the majority of apneas are due instead to prematurity, and only a few apneas are associated with reflux events. Furthermore, Corvaglia et al (16) more recently found a TA in a similar population of premature infants at <32 weeks by using a PSG and combined MII-pH probe.

Di Fiore et al (15) excluded a TA between reflux and apneas in infants slightly younger than our patients. However, the infants were recruited based on the presence of clinical GER with or without persistent CR events, whereas we recruited only patients with persistent CR events at term. Furthermore, their study was limited by the fact that they did not use an MII-pH probe excluding pH >4 reflux from the analysis. The number of OAs may also have been underestimated by not using a flow sensor and a PSG.

Mousa et al (13), who studied a different population of older infants with a history of apparent life-threatening events, found results similar to those in our present study. By using a χ2 analysis, they were able to demonstrate a TA only in particular infants.

Back to Top | Article Outline

Window of Association

The window of association is the length of time following a reflux event that an apneic event is considered to be associated with that reflux event. An optimal size for that association window has not been defined, with the etiology of the reflux-apnea relation still unknown. The mechanism of reflux-induced apneas is most likely multifactorial, secondary to an esophageal distention induced by GER (25) or stimulation of a chemolaryngeal reflex protecting the airway from aspiration (26,27). For the first mechanism, the apnea should occur simultaneously with the reflux event; but for the second mechanism, the apnea may be simultaneous or delayed as liquid accumulates in the oropharynx and eventually overwhelms the capacity of airway protection. We speculate that the capacities of airway protection may vary as a function of the volume of the refluxate; the degree of acidity of the reflux event, which may damage the laryngeal adductor reflex (3,4); and the degree of neuronal maturity and integrity.

In a novel approach to more fully explore the effect of the size of the association window, we varied the association window sizes for both SI and Fisher exact test analyses. For the SI with OA, we looked at multiple windows of association that varied between 15 and 300 seconds. The SAP was designed with a 2-minute association window and bin size. Our results from the Fisher exact test showed a TA as early as 15 to 45 seconds, with the lowest P value around 120 seconds, suggesting that, in this limited set of patients, the optimal window of association is around 120 seconds, the same as the value used in Sandhill's SAP measure. We also found in our patients that a significant TA between OA and MII-pH ≤4 occurred earlier than for MII-pH >4. Varying the temporal window size produced fluctuations in the P value that can vary more than or less than the traditional thresholds of P = 0.05. These results demonstrate the limitations of the artificial concept of binning used in the Fisher exact test by the exclusion of reflux or OA events if more than 1 event occurs within the same bin. Furthermore, it demonstrates the importance of looking at a range of TA windows to characterize TA. Looking at only 1 window of association may not be enough to characterize TA. Additional studies should continue to focus on determining the optimal window of association and comorbidity factors (laryngeal sensation, brain abnormalities) that could trigger the GER-induced apneas.

Back to Top | Article Outline

Effect of the Presence of Nasal Probe on the Risk of Increased CR Events

The presence of a nasal MII-pH probe was evaluated as potentially increasing the risk of nasal obstruction and OA. For that reason, we removed the probe and continued the PSG alone for 3 hours. The frequency of OA was unchanged after the removal of the MII-pH probe. In fact, according to previous studies, the presence of a nasal probe may actually decrease the risk of OA in adults and infants (28–33) by opposing mucosal adhesion forces, decreasing the collapsibility of the pharynx, separating the tongue from the posterior pharyngeal wall, or increasing swallowing frequency resulting in a higher tone of the pharynx-dilating muscles. To address the potential risk of nasal obstruction, we also removed the feeding tube after each feeding.

Back to Top | Article Outline


TA between CR events and reflux events has been demonstrated in a single-subject analysis in former premature infants with persistent CR events at term. Criteria for fundoplication, based on clinical history and pH probe alone, together with the respective outcomes, were supported by the results of the TA, suggesting that this method may be useful as a clinical diagnostic tool before considering a fundoplication. Additional studies are needed to confirm the relation between TA and outcome. Additionally, pH-only events may have only a limited role in generating CR events. In our study, we generally found a window size of approximately 120 seconds (reflecting the association window size used in previous studies), but the physiological basis for the length of this association window still needs to be understood.

Back to Top | Article Outline


Statistical advice for the present study was provided by Michael Griswold, PhD, associate professor, adjunct executive director, Johns Hopkins Biostatistics Center, Baltimore, MD; and Gang Chen, PhD, mathematical statistician, Scientific and Statistical Computing Core, Intramural Research Program at NIMH, NIH, Bethesda MD.

Back to Top | Article Outline


1. Ramanathan R, Corwin MJ, Hunt CE, et al. Cardiorespiratory events recorded on home monitors: comparison of healthy infants with those at increased risk for SIDS. JAMA 2001; 285:2199–2207.
2. Hunt CE, Corwin MJ, Baird T, et al. Cardiorespiratory events detected by home memory monitoring and one-year neurodevelopmental outcome. J Pediatr 2004; 145:465–471.
3. Suskind DL, Thompson DM, Gulati M, et al. Improved infant swallowing after gastroesophageal reflux disease treatment: a function of improved laryngeal sensation? Laryngoscope 2006; 116:1397–1403.
4. Thompson DM, Rutter MJ, Rudolph CD, et al. Altered laryngeal sensation: a potential cause of apnea of infancy. Ann Otol Rhinol Laryngol 2005; 114:258–263.
5. Cristofalo E, Nunez J, Mathias K, et al. Duration of apnea and bradycardia predicts oral feeding success in premature infants 30 weeks gestation. Paper presented at: APS-SPR; 2006; San Francisco, CA.
6. Lasser MS, Liao JG, Burd RS. National trends in the use of antireflux procedures for children. Pediatrics 2006; 118:1828–1835.
7. Kapoor S, Shashidhar H, Iocono J, et al. Fundoplication: a comparative study in extremely low birth weight and term and near term infants. PAS 2008. Hawaii 4455.26.
8. Wenzl TG, Schenke S, Peschgens T, et al. Association of apnea and nonacid gastroesophageal reflux in infants: investigations with the intraluminal impedance technique. Pediatr Pulmonol 2001; 31:144–149.
9. Rosen R, Nurko S. The importance of multichannel intraluminal impedance in the evaluation of children with persistent respiratory symptoms. Am J Gastroenterol 2004; 99:2452–2458.
10. Condino AA, Sondheimer J, Pan Z, et al. Evaluation of gastroesophageal reflux in pediatric patients with asthma using impedance-pH monitoring. J Pediatr 2006; 149:216–219.
11. Thilmany C, Beck-Ripp J, Griese M. Acid and non-acid gastro-esophageal refluxes in children with chronic pulmonary diseases. Respir Med 2007; 101:969–976.
12. Peter CS, Sprodowski N, Bohnhorst B, et al. Gastroesophageal reflux and apnea of prematurity: no temporal relationship. Pediatrics 2002; 109:8–11.
13. Mousa H, Woodley FW, Metheney M, et al. Testing the association between gastroesophageal reflux and apnea in infants. J Pediatr Gastroenterol Nutr 2005; 41:169–177.
14. Magistà AM, Indrio F, Baldassarre M, et al. Multichannel intraluminal impedance to detect relationship between gastroesophageal reflux and apnoea of prematurity. Dig Liver Dis 2007; 39:216–221.
15. Di Fiore JM, Arko MK, Whitehouse M, et al. Apnea is not prolonged by acid gastroesophageal reflux in preterm infants. Pediatrics 2005; 116:1059–1063.
16. Corvaglia L, Zama D, Gualdi S, et al. Gastroesophageal reflux increases the number of apneas in very preterm infants. Arch Dis Child Fetal Neonatal Ed 2009; 94:F188–F192.
17. Nunez J. Extra-esophageal manifestations of gastroesophageal reflux in the premature infant: new clinical methods of evaluation. eNeonatal Review 2007;5. http://hopkinscme.edu/ofp/eneonatalreview/newsletters/1107.html.
18. Wenzl TG, Moroder C, Trachterna M, et al. Esophageal pH monitoring and impedance measurement: a comparison of two diagnostic tests for gastroesophageal reflux. J Pediatr Gastroenterol Nutr 2002; 34:519–523.
19. Peter CS, Wiechers C, Bohnhorst B, et al. Detection of small bolus volumes using multiple intraluminal impedance in preterm infants. J Pediatr Gastroenterol Nutr 2003; 36:381–384.
20. Frieling T, Hermann S, Kuhlbusch R, et al. Comparison between intraluminal multiple electric impedance measurement and manometry in the human oesophagus. Neurogastroenterol Motil 1996; 8:45–50.
21. Iber C, Ancoli-Israel S, Chesson A, et al. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. Westchester, IL: American Academy of Sleep Medicine; 2007.
22. Bredenoord AJ, Weusten BL, Smout AJ. Symptom association analysis in ambulatory gastro-oesophageal reflux monitoring. Gut 2005; 54:1810–1817.
23. Woodley F, Mousa H. pH-only acid reflux events in infants during later phases of the feeding cycle are less acidic and cleared more efficiently than classic 2-phase acid reflux events. J Pediatr Gastroenterol Nutr 2009; 48:41–47.
24. Woodley F, Mousa H. Acid gastroesophageal reflux reports in infants: a comparison of esophageal pH monitoring and multichannel intraluminal impedance measurements. Dig Dis Sci 2006; 51:1910–1916.
25. Jadcherla SR. Upstream effects of esophageal distension: effect on airway. Curr Gastr Rep 2006; 8:189–193.
26. Thach BT. Maturation and transformation of reflexes that protect the laryngeal airway from liquid aspiration from fetal to adult life. Am J Med 2001; 111:69s–77s.
27. Pickens DL, Schefft G, Thach BT. Prolonged apnea associated with upper protective reflexes in apnea of prematurity. Am Rev Respir Dis 1988; 137:113–118.
28. Karlson KH Jr, Chaudary BA, Porubsky ES. Long term nasopharyngeal intubation in obstructive sleep apnea. Pediatr Pulmonol 1987; 3:440–442.
29. Aubert G. Alternative therapeutic approaches in sleep apnea syndrome. Sleep 1992; 15:S69–S72.
30. Groswasser J, Scaillon M, Rebuffat E, et al. Naso-oesophageal probes decrease the frequency of sleep apnoeas in infants. J Sleep Res 2000;9:193–6.
31. Series F, Cormier Y, Laforge J, et al. Mechanisms of the effectiveness of continuous positive airway pressure in obstructive sleep apnea. Sleep 1992; 15:S47–S49.
32. Heaf DP, Helms PJ, Dinwiddie R, et al. Nasopharyngeal airways in Pierre Robin syndrome. J Pediatr 1982; 100:698–703.
33. Strohl KP, Redline S. Nasal CPAP therapy. Upper airway muscle activation, and obstructive sleep apnea. Am Rev Respir Dis 1986; 134:555–558.

cardiorespiratory events; gastroesophageal reflux; multichannel intraluminal impedance; obstructive sleep apnea; pH probe; polysomnography; premature infants

Copyright 2011 by ESPGHAN and NASPGHAN