Regurgitation is a common phenomenon during the first year of life. At least two thirds of 4-month-old and 5% of 12-month-old infants have regurgitation or vomiting. An indefinite number of subjects may have less obvious manifestations of gastroesophageal reflux disease (GERD) such as persistent crying, back-arching, abnormal posturing, poor growth, chronic respiratory manifestations and sleeping and feeding problems (1-3) Many, but not all, infants with GERD have frequent regurgitation. GERD should be suspected if the regurgitating infant has one or more other symptoms such as crying, fussing or arching back, refusal to feed, failure to thrive or hematemesis. However, most of these phenomena (except failure to thrive and hematemesis) occur in healthy infants (4,5). Hence, discrimination between healthy and GERD on a clinical basis is difficult. The ability to discriminate is essential in determining which infants to submit to investigations and treatment.
In 1993 and 1996, Orenstein et al. proposed a clinical questionnaire (I-GERQ) as a simple tool to identify children with GERD (2,6). Since other authors have not subsequently use the suggested score, the predictability of GERD in infants based on clinical history is still unclear.
The aims of this study were to determine the prevalence of the clinical manifestations of reflux in normal infants and to characterize the diagnostic validity of the I-GERQ and of a modified questionnaire for separating normal infants from those with GERD.
MATERIALS AND METHODS
We recruited 200 infants (age, 0.5-12 months; median age, 4 months) referred to our hospital: 100 were referred because of symptoms suggesting GERD and 100 attended a well-baby clinic. The selection criteria to include infants at the well-baby clinic were absence of any known disease or any medical/dietary treatment during the 2 weeks preceding the questionnaire and absence of concern by parents or family doctor about the well being of the baby. No investigation was performed in the infants presenting at the well-baby clinic for ethical reasons. All parents completed a 35-item (Orenstein's modified) questionnaire regarding characteristics of regurgitation, feeding, bowel habit, crying, hiccups, respiratory symptoms, weight gain, family history of GER and allergy and parental suspicion of GERD. We added 18 questions to the original Orenstein questionnaire: one about projectile regurgitation, one about difficulties in burping, one about noisy respiration, three about pneumonia, bronchitis or chronic cough, two about frequency and duration of hiccups, five about bowel habits, four about family history of reflux disease and allergy and one about parental suspicion of reflux disease in their infant (Table 1). We applied the I-GERQ GERD score for 10 items (17 questions) from our questionnaire (Table 1) of the 11 original Orenstein-scored items (2), as no specific question about spells of back arching was investigated in our patients. The question was omitted because adequate translation was difficult. The questions were translated into French and Dutch and re-translated in English. This question scored 2 points (on a total maximal possible score of 25 points) in Orenstein's score and because the proposed best cut-off score was >7 points (positive predictive value of 1.00 and negative predictive value of 0.98), we considered a score for the I-GERQ of <6 as negative. A score of >7 was considered positive, and the score was borderline if it was 6 or 7. The questionnaire was filled in by one of the parents, who read and marked it without assistance. For statistical analysis we considered only positive or negative answers, leaving the original results without interpretation of missing answers. The validation of our questionnaire was based on test-retest (intra-observer) consistency of each question in 15 randomly selected parents, resubmitting the questionnaire to the same responder parent by phone interview made by the same nurse of the unit 1 week later.
All 100 infants referred because of GER symptoms (regurgitation or vomiting with or without distress) had a 24-hour pH study. In 44 an upper endoscopy with esophageal biopsy was also performed. The pH study was performed with a Mark III Digitrapper (Synectics Medical, Stockholm, Sweden) connected to an antimony electrode (Single Cristal Antimony Multi-Use pH Catheter, Synectics Medical). The electrode was calibrated before and after each investigation. The location of the electrode was verified by fluoroscopy at the third vertebra above the diaphragm. Data were analyzed with the EsopHagram software from Gastrosoft, using Polygram for Windows. The reflux index (RI, the duration that the recorded pH was <4.0, expressed as a percentage of the total duration of the pH monitoring) was calculated for each patient, and the cut-off for "normal" versus "abnormal" was set at 10%. Endoscopy was proposed to all parents, but only 44 accepted. The endoscopy was performed before the pH monitoring. The infants were sedated with midazolam intrarectally. Esophageal grasp biopsies were taken at approximately 3 cm above the cardia. A diagnosis of esophagitis was made if any of the following criteria were met: basal zone hyperplasia (>20%) and papillary lengthening (>50%) or the presence of many (≥7) neutrophils, lymphocytes or at least a few (4-6 per biopsy) eosinophils in the esophageal epithelium, lamina propria or both. Minimal abnormal biopsy was defined in presence of mildly increased inflammatory infiltration in the esophageal mucosa (7-9).
We used χ2, logistic and multiple regression analysis for statistical purpose. Logistic and multiple regression analyses were applied to find a correlation among answers (and combination of answers) and results of pH study (RI >5% and RI >10%) and histology (normal, minimal abnormality, esophagitis). The receiver operating characteristic curve was applied based on different cut-off (numbers) of total positive answers in the questionnaire and positive pH study results. The evaluation of numbers of positive answers in subjects with a pathologic pH monitoring (considered as "gold standard") was necessary to define true positives and false positive rate for the different cut-off points (10). The Hospital Ethics Committee approved the study.
Symptoms in healthy infants and infants suspected of GERD
Healthy infants had a high prevalence of reflux symptoms such as daily regurgitation (45%), crying more than 1 hour per day (20%), and daily hiccups (35%). The infants referred for GERD had significantly more regurgitation (68%; P < 0.05) and crying (51%; P < 0.01). Hiccups were more frequent (49%) but the difference did not reach statistical significance.
Validation of the Questionnaire and Statistical Analysis
The median test-retest consistency for the 35 items was 0.9 (mean, 0.96; SD, 0.5; range, 0.8 to 1.0). Seventy percent of the questionnaires were complete, whereas 30% showed missing answers. The mean number of blank (missing) answers was 0.9 (range, 0 to 8). Missing answers are shown in Table 2. In patients with a pathologic pH-metry, only 12 questions among the total of 35 of the questionnaire showed a majority of positive answers. Plotting the true positive rate against the false positive rate for the 13 different cut-off points (from 0 to 12), a ROC curve was not obtained (10). Logistic and multiple regression analysis were not significant.
pH-Monitor and Questionnaire Results in Infants Suspected of GERD
The pH study was considered pathologic (RI > 10%) in 21 of 100 (21%). Abnormal pH monitoring was significantly associated with pneumonia or apnea with fussing (P = 0.013 for both), defecation less than once a day (P = 0.033) and constipation (P = 0.05) (Table 2).
Histology and pH Monitoring in GERD Suspected Infants
In infants undergoing endoscopy, histologic esophagitis was present in 17 of 44 (39%). Not one question was significantly predictive for the presence of esophagitis. No meaningful correlation could be found between the results of the pH study and histology (38% of infants with a pathologic pH study had a normal biopsy and 53% of the infants with abnormal histology of esophageal biopsies had a normal pH study). Discordance between pH study and biopsies was found in 14 of 44 (32%) patients. Lowering the pathologic cut-off of pH study to 5% (RI >5% considered as pathologic), 35% of the infants with abnormal biopsies had still a normal pH study and 48% of the infants with a pathologic pH study had still normal biopsies (Table 3). For prediction of histologic esophagitis, sensitivity and specificity were 47% and 81% for a RI >10% and 65% and 63% for a RI >5%, respectively. The positive predictive value and negative predictive value were 0.62 and 0.71 for a pathologic RI set at >10% and 0.52 and 0.74 for a pathologic RI set at >5%, respectively.
Orenstein Score, Histology and pH Study in Infants with Suspected GERD
In our study, five of 17 (29%) patients with abnormal biopsies and four of 21 (19%) patients with pathologic pH study had an Orenstein score <6 points (only one patient had both abnormal biopsy and pathologic pH study). Overall, the defined cut-off of >7 failed to identify eight of 31 (26%) infants with GERD. Conversely, the score was positive in 17 of 22 (81%) infants with both normal biopsy and pH study and in 14 of 47 (30%) infants with normal pH study (not biopsied). Lowering the pathologic reflux index to 5%, we still found 12 of 17 (71%) patients with a positive clinical score (>7) and both normal histology and pH study. One patient with normal score but with pathologic pH study had frequent and copious regurgitation without any other symptoms. Two patients with esophagitis had a RI <5% and a clinical score of less than 6.
In 18 of 100 (18%) patients we observed a borderline score of 6 - 7 points. Results of pH-metry and histology of esophageal biopsies were available in nine of 18 children. Three had esophagitis (two of three with a normal pH study) and six has normal histology (three of the six had aabnormal pH-metry). In the other nine infants only the pH-metry result was available, and was abnormal in only one infant.
The discrimination between physiologic and pathologic GER and its manifestations remains a challenge in infants. In most infants, regurgitation has a benign prognosis with spontaneous resolution by 12-18 months. Nonetheless, troublesome GERD is reported in 5% to 8% of infants (11-14), ranging from esophageal mucosal damage (esophagitis, Barrett's and esophageal stenosis) to extra-esophageal complications (feeding and sleeping disturbances, weight loss, chronic respiratory symptoms and impaired quality of life). Feeding refusal may also represent a long-term sequela of infantile chronic reflux (15) resulting from a persistent memory of difficult feeding (15). Lack of medical referral, self-treatment and heterogenicity of diagnostic criteria in epidemiologic data, make it impossible to establish the exact prevalence of GER and GERD, particularly considering if the manifestations are extra-esophageal.
Recent recommendations suggest that clinical history may be a sufficient and reliable tool to diagnose GERD (16). It is suggested that the absence of persistent crying, feeding problems, poor weight gain or respiratory problems should exclude esophagitis or pathologic pH study (16). However, this opinion is not substantiated by evidence-based data in infants. We cannot confirm a clinically relevant correlation between clinical history and Orenstein score in the prediction of abnormal pH-metry or esophagitis. In our questionnaire, we added items to cover the wide spectrum of clinical presentations of GERD. As we modified the original validated Orenstein questionnaire, we validated our questionnaire by an intra-observer consistency in a small randomly selected sample of parents. Unfortunately, in our experience, no symptom was significantly associated with esophagitis and only pneumonia, apnea and constipation were significantly associated with a pathologic result on pH monitoring. We also failed to identify a clinical symptom or complaint that would distinguish abnormal from normal pH-metry and esophagitis from normal histology.
According to literature, the incidence of histologic esophagitis in symptomatic children ranges from 15% to 62% (17-20) and is not directly related to the duration of symptoms.
Defining the criteria to distinguish GER from GERD is particularly important to avoid over-diagnosis and over treatment or delayed diagnosis and its complications. In adults, a detailed clinical history and a specific questionnaire seem an accurate instrument to predict pathologic reflux. In a recent study of 240 adults, a questionnaire based on the presence and frequency of seven symptoms (heartburn, regurgitation, dysphagia, chest pain, nocturnal cough, dysphonia and asthma) resulted in a sensitivity, specificity and positive predictive value for GERD of greater than 91% (21).
The usefulness of a detailed clinical history in infants was first analyzed by Orenstein et al. (2,6) through a validated infant GER questionnaire (I-GERQ) with 26 questions administered to the parents of 100 normal subjects and only 35 infants with a positive pH study (RI >10%) or histologic esophagitis. The group of presumed (not investigated) normal infants had a high prevalence of symptoms suggesting GER but from a 25-point GERD score based on 11 items, a score cut-off of >7 resulted in a high positive and negative predictive value (positive predictive value of 1.00 and negative predictive value of 0.98) separating infants with physiologic from those with pathologic reflux (2,6). More recently, 125 infants with persistent distress and symptoms suggesting GERD were retrospectively evaluated by pH study and esophageal biopsy (22). Esophagitis was detected in 26% of cases (in only 28% of them was it associated with RI >10%), whereas 18% presented abnormal pH study but normal histology. This study showed a significant correlation between regurgitation more than five times per day and an abnormal reflux index but no correlation between esophagitis and symptoms (regurgitation, hematemesis, feeding difficulties, atopic dermatitis and failure to thrive). It also confirms a poor diagnostic agreement between histology and pH study in infants. The concordance between pH study and histology has been debated, denied or sustained in previous reports (8,9,17,18,23-29).
In three other pediatric studies of persisting GER symptoms, no relation could be demonstrated between symptoms (no specified questionnaire used) and esophagitis (8,17,30). Conversely, one study investigating 40 children (mean age, 79.6 ± 5l.9 months) with GERD and seven symptomatic children (mean age, 52.6 ± 37.0 months) without GERD, showed a good correlation between histology and clinical score, although immunohistochemical markers used (CD3, CD25, ICAM, HLADR) were not predictive for symptom severity (31).
Several issues arise from our study. We can confirm the common occurrence of GER-related symptoms in an unselected healthy infant population. Only pneumonia, apnea with fussing and constipation were significantly associated with pathologic pH probe results. Whether these are findings are a coincidence or clinically meaningful remains open for debate. Clinical history was unreliable to predict esophagitis. We could not confirm the predictability of Orenstein's clinical score, as it failed to identify 26% of our patients with esophagitis or pathologic pH study and was falsely positive in as many as 81% of the infants with both normal pH monitoring and histology. These results reiterate the low agreement/affinity between pH-metry and histology. The lack of correlation between clinical history, pH study and esophageal histology may be ascribed to the selected cut-off defining pathologic pH study, the suboptimal sensitivity of pH study in infants, the differences in esophageal mucosal resistance and esophageal hypersensitivity, the parental perception and report of symptoms and possible genetic factors (18). Intraluminal impedance analysis could possibly better clarify in the future the clinical or histologic correlation adding more information about non-acid reflux that may play a critical role, especially in infants (32).
Infants suspected of GERD have more frequent regurgitation, vomiting and crying than healthy control infants. Questionnaires can help in separating healthy infant from infants with GERD. However, on the basis of our findings, clinical history and questionnaires cannot predict the severity of GERD. Because the severity of GERD is largely unpredictable in infants, a highly sensitive and specific method to select infants for empiric pharmacotherapy and a cost-benefit investigation program still needs to be developed. Normal pH study or normal histology does not exclude GERD, and accurate clinical history, pH study and endoscopy are complimentary diagnostic tools.
1. Chouhou D, Rossignol C, Bernard F, Dupont C. Le reflux gastro-oesophagien. dans le centres de bilan de santé de l'enfant de moins de 4 ans. Arch French Pediatr
2. Orenstein SR, Shalaby TM, Cohn J. Reflux symptoms in 100 normal infants: diagnostic validity of the Infant Gastroesophageal Reflux Questionnaire. Clin Pediatr
3. Nelson SP, Chen EH, Syniar GM, Christoffel KK. Prevalence of symptoms of gastroesophageal reflux in infancy. Arch Pediatr Adolesc Med
4. Heine RG, Jaquiery A, Lubitz L, Cameron DJ, Catto-Smith AG. Role of gastro-oesophageal reflux in infant irritability. Arch Dis Child
5. Snel A, Barnett CP, Cresp TL, Haslam RR, Davidson GP, Malbert TH, Dent J, Omari TI. Behavior and gastroesophageal reflux in the premature neonate. J Pediatr Gastroenterol Nutr
6. Orenstein SR, Cohn JF, Shalaby TM, Kartan R. Reliability and validity of an infant gastroesophageal questionnaire. Clin Pediatr
7. Vandenplas Y. Reflux esophagitis in infants and children. A report from the Working Group of the European Society of Pediatric Gastroenterology and Nutrition on Gastro-oesophageal Reflux Disease. J Pediatr Gastroenterol Nutr
8. Hyams JS, Ricci A, Leichtner AM. Clinical and laboratory correlates of esophagitis in young children. J Pediatr Gastroenterol Nutr
9. Black DD, Haggitt RC, Orenstein SR. Esophagitis in infants. Morphometric histological diagnosis and correlation with measures of gastroesophageal reflux. Gastroenterology
10. Metz C.E. Basic principles of ROC analysis. Semin Nucl Med
11. Wienbeck M, Barnert J. Epidemiology of reflux disease and reflux esophagitis. Scand J Gastroenterol
12. Carrè I. The natural history of the partial thoracic stomach ("hiatal hernia") in children. Arch Dis Child
13. Sacre L, Vandenplas Y. Gastroesophageal reflux associated with respiratory abnormalities during sleep. J Pediatr Gastroenterol Nutr
14. Vandenplas Y, Goyvaerts H, Helven R, Sacre L..Gastroesophageal reflux, as measured by 24-hours pH-monitoring, in 509 healthy infants screened for risk of sudden infants death syndrome. Pediatrics
15. Nelson SP, Chen EH, Syniar GM, Christoffel KK. One year follow-up of symptoms of gastroesophageal reflux during infancy. Pediatrics
16. Rudolph CD, Mazur LJ, Liptak GS, et al. Guidelines for evaluation and treatment of gastroesophageal reflux in infants and children. Recommendations of the North American Society for Pediatric Gastroenterology and Nutrition. J Pediatr Gastroenterol Nutr.
17. Treem W, Davis P, Hyams J. Gastroesophageal reflux in the older child: presentation, response to treatment, and long-term follow-up. Clin Pediatr
18. Lee WS, Beattie RM, Meadows N, Walker-Smith JA. Gastro-oesophageal reflux: clinical profiles and outcome. J Paed Child Health
19. Tolaymat N, Chapman DM. Gastroesophageal reflux disease in children older than two years of age. W V Med J
20. El-Serag HB, Bailey NR, Gilger M, Rabeneck L. Endoscopic manifestations of gastroesophageal reflux disease in patients between 18 months and 25 years without neurological deficits. Am J Gastroenterol
21. Manterola C, Munoz S, Grande L, Bustos L. Initial validation of a questionnaire for detecting gastroesophageal reflux disease in epidemiological settings. J Clin Epidemiol
22. Heine RG, Cameron DJ, Chow CW, Hill DJ, Catto-Smith AG. Esophagitis in distressed infants: poor diagnostic agreement between esophageal pH monitoring and histopathologic findings. J Pediatr
23. Vandenplas Y, Franckx-Goossens A, Pipeleers-Marichal M, Derde MP, Sacre-Smits L. Area under pH 4: advantages of a new parameter in the interpretation of esophageal pH monitoring data in infants. J Pediatr Gastroenterol Nutr
24. Biller JA, Winter HS, Grand RJ, Allred EN. Are endoscopic changes predictive of histologic esophagitis in children? J Pediatr
25. Cucchiara S, Staiano A, Gobio Casali L, Boccieri A, Paone FM. Value of the 24 hour intraoesophageal pH monitoring in children. Gut
26. Winter HS, Madara JL, Stafford RJ, et al. Intraepithelial eosinophils: a new diagnostic criterion for reflux esophagitis. Gastroenterology
27. Avidan B, Sonnenberg A, Schnell TG, Sontag SJ. Acid reflux is a poor predictor for severity of erosive reflux esophagitis. Dig Dis Sci
28. Stal P, Lindberg G, Ost A, Iwarzon M, Seensalu R. Gastroesophageal reflux in healthy subjects. Significance of endoscopic findings, histology, age, and sex. Scand J Gastroenterol.
29. Ashorn M, Ruuska T, Karikoski R, Laippala P. The natural course of gastroesophageal reflux disease in children. Scand J Gastroenterol
30. Chadwick LM, Kurinczuk JJ, Hallam LA, Brennan BA, Forbes D. Clinical and endoscopic predictors of histological oesophagitis in infants. J Paediatr Child Health
31. Tozzi A, Staiano A, Paparo F, et al. Characterization of the inflammatory infiltrate in peptic oesophagitis. Dig Liver Dis
32. Wenzl TG. Investigating esophageal reflux with the intraluminal impedance technique. J Pediatr Gastroenterol Nutr