Overlapping symptoms of gastroesophageal reflux and aspiration highlight the limitations of validated questionnaires

Abstract

Objectives:

Infants frequently present with feeding difficulties and respiratory symptoms, which are often attributed to gastroesophageal reflux but may be due to oropharyngeal dysphagia with aspiration. The Infant Gastroesophageal Reflux Questionnaire Revised (I-GERQ-R) is a clinical measure of gastroesophageal reflux disease but now there is greater understanding of dysphagia as a reflux mimic. We aimed to determine the degree of overlap between I-GERQ-R and evidence of dysphagia, measured by Pediatric Eating Assessment Tool-10 (Pedi-EAT-10) and videofluoroscopic swallow study (VFSS).

Methods:

We performed a prospective study of subjects <18 months old with feeding difficulties. All parents completed Pedi-EAT-10 and I-GERQ-R as a quality initiative to address parental feeding concerns. I-GERQ-R results were compared to Pedi-EAT-10 and, when available, results of prior VFSS. Pearson correlation coefficients were calculated to determine the relationship between scores. Groups were compared with one-way ANOVA and Fisher’s exact test. ROC analysis was completed to compare scores with VFSS results.

Results:

108 subjects with mean age 7.1±0.5 months were included. Pedi-EAT-10 and I-GERQ-R were correlated (r=0.218, p=0.023) in all subjects and highly correlated in the 77 subjects who had prior VFSS (r=0.369, p=0.001). The blue spell questions on I-GERQ-R had relative risk 1.148 (95% CI 1.043–1.264, p=0.142) for predicting aspiration/penetration on VFSS, with 100% specificity. Scores on the question regarding crying during/after feedings were also higher in subjects with abnormal VFSS (1.1±0.15 vs 0.53±0.22, p=0.04).

Conclusions:

I-GERQ-R and the Pedi-EAT-10 are highly correlated. I-GERQ-R results may actually reflect oropharyngeal dysphagia and not just gastroesophageal reflux disease in infants.

Introduction

Infants and young children frequently present to gastroenterologists for evaluation of feeding difficulties and respiratory symptoms. These troublesome symptoms, ranging in severity from spit-ups to choking to cyanotic spells, are often attributed to gastroesophageal reflux but may be due to oropharyngeal dysphagia with aspiration(1–5). Symptom misattribution by parents and providers not only leads to an incorrect diagnosis being made, it can also result in inappropriate medical management, including use of potentially harmful acid-suppressing medications in infants who do not need acid-suppression, risk of unaddressed pulmonary damage from ongoing aspiration, and longer-term risks to growth, nutrition, and oral aversion(6–10).

Both reflux and aspiration can be difficult to diagnose and manage without objective testing. Oropharyngeal dysphagia with aspiration is frequently silent in infants and therefore requires a high index of suspicion and videofluoroscopic swallow study (VFSS) testing to confirm the diagnosis(3, 11, 12). Gastroesophageal reflux is common in both well and medically-complex infants; therefore, many symptoms are attributed to reflux based on symptoms alone, but objective testing to prove association of reflux events to these symptoms requires pH/impedance testing. Because these tests require additional trips to the hospital, can be costly and require patient participation, symptom-based questionnaires to aid in diagnosis have been developed(13–15).

The Infant Gastroesophageal Reflux Questionnaire Revised (I-GERQ-R) is a 12-item questionnaire developed in 2006 as a validated measure of infant gastroesophageal reflux disease (GERD)(16). It has historically been used to both diagnose and monitor reflux symptom burden in epidemiologic and pivotal pharmacologic studies(17–23). With an abnormal cutoff above a score of 16 as signifying clinically significant GERD as validated against caregiver symptom reports, I-GERQ-R continues to be widely utilized for studies of reflux prevalence and clinical trial reporting(16, 24–29).

The Pediatric Eating Assessment Tool-10 (Pedi-EAT-10) was more recently developed as a parent-reported measure of feeding difficulties. This 10-item questionnaire was developed in 2017 from a similar assessment used in adults and was validated against penetration-aspiration scale (PAS) scores on VFSS(30–34). Pedi-EAT-10 has been proposed to predict swallowing dysfunction and aspiration in a variety of pediatric patients, including children with esophageal atresia and children with neurologic impairments, with abnormal Pedi-EAT-10 values reported as scores of 4, 10 or 12, depending on the population studied(32, 33, 35).

While I-GERQ-R and Pedi-EAT-10 are considered robust instruments for diagnosis and disease monitoring(32), it is not known whether there is overlap between these two instruments and to what degree I-GERQ-R is capturing symptoms of aspiration. The primary aim of this study was to evaluate the association between I-GERQ-R and Pedi-EAT-10 scores in a prospective cohort of young children presenting with feeding difficulties and respiratory symptoms. We hypothesized that I-GERQ-R and Pedi-EAT-10 scores are correlated due to overlapping symptoms of reflux and oropharyngeal dysphagia and that I-GERQ-R may be capturing aspiration symptoms rather than GERD alone.

Methods

We retrospectively reviewed Pedi-EAT-10 and I-GERQ-R questionnaires distributed as part of a quality initiative to identify the frequency of feeding difficulties in patients in an effort to address deficiencies in feeding supports for families. Between October 2018 and March 2020, all families of patients under 18 months of age presenting to the Aerodigestive Center at Boston Children’s Hospital were approached to complete questionnaires as part of a quality initiative to assess frequency of feeding difficulty in the multidisciplinary clinic.

All study participants’ parents completed both Pedi-EAT-10 and I-GERQ-R, as previously described(16, 32, 33, 35). Instrument questions are shown in the Table, Supplemental Digital Content. Pedi-EAT-10 scores greater than 12 and I-GERQ-R scores greater than or equal to 16 were considered abnormal for group comparisons. Parents also completed a survey of medications including the use of proton pump inhibitors (PPI), H2 receptor antagonists (H2RA), and erythromycin, dietary restrictions including dairy/soy elimination or hypoallergenic formula, and thickening status at time of questionnaire completion.

A subset of subjects had a clinically indicated VFSS performed and were classified as having aspiration or normal swallow function. VFSS showing aspiration or penetration were considered abnormal based on prior studies(36–39). Medical records were reviewed for demographics and additional clinical data including medications and diagnostic study results.

One of the aims of this study was to determine the correlation between Pedi-EAT-10 and I-GERQ-R as a measure of overlapping symptoms of oropharyngeal dysphagia and gastroesophageal reflux. Pearson correlation coefficients were calculated to determine the relationship between Pedi-EAT-10 and total I-GERQ-R scores in the cohort overall and in the sub-group of subjects with VFSS results. Fisher’s exact test was performed to compare groups, including those with normal/abnormal VFSS, normal/abnormal Pedi-EAT-10, and those with normal/abnormal I-GERQ-R scores. Student’s t-test was used to compare mean scores. One-way ANOVA was used to evaluate each questionnaire item’s discriminant ability to predict an abnormal score on the other scale and abnormal VFSS results.

To confirm the validity of Pedi-EAT-10 to predict aspiration in our cohort, we performed ROC analysis to evaluate the relationship between Pedi-EAT-10 scores and VFSS results. Sensitivity and specificity were calculated for Pedi-EAT-10 cutoffs based on prior published values(32, 33, 35). To further evaluate the relationship between I-GERQ-R scores and dysphagia, we also performed an ROC analysis between I-GERQ-R scores and VFSS results.

Data are presented as mean±SE and % (n) unless indicated otherwise. All statistical tests were 2-sided with p-value <0.05 considered statistically significant. All data were analyzed using SPSS Statistics (version 23). The present study was approved by the Institutional Review Board at Boston Children’s Hospital.

Results

108 subjects with mean age of 7.06±0.50 months were included in the full cohort analysis. 50% were female. All subjects completed both Pedi-EAT-10 and I-GERQ-R questionnaires. Subjects had mean I-GERQ-R score of 13.01±0.63 with 36% (39) having abnormal scores. Subjects had mean Pedi-EAT-10 score of 8.7±0.80 with 30% (32) having abnormal scores. For the cohort overall, the Pearson correlation between the two scores was 0.218 (p=0.023).

Table 1 shows the comparison between those with normal and abnormal I-GERQ-R and is notable for higher mean Pedi-EAT-10 in those with abnormal I-GERQ-R. There was no difference in proportion with abnormal I-GERQ-R depending on PPI, H2RA, any acid suppression, erythromycin, hypoallergenic formula, thickening, or enteral tube presence. Of the 53 subjects on anti-reflux therapies, 63% (n=33) were reported to have reflux symptoms (spit-up, arching/fussiness, feeding intolerance) by their gastroenterologist. There was no significant association between physician report of reflux symptoms and abnormal I-GERQ-R (p=0.375).

Table 1:

Comparison Between Subjects with Normal Vs Abnormal I-GERQ-R Results

	I-GERQ-R Abnormal (n=39)	I-GERQ-R Normal (n=69)	P-value
Mean Age (months)	5.84 ± 0.76	7.75 ± 0.65	0.058
VFSS Performed	69% (27)	74% (51)	0.658
VFSS Abnormal	85% (23/27)	78% (39/50)	0.555
Pedi-EAT-10 Abnormal	39% (15)	25% (17)	0.188
Mean Pedi-EAT-10 Score	11.40 ± 1.41	7.18 ± 0.92	0.015
I-GERQ-R Abnormal	100% (39)	0% (0)
Mean I-GERQ-R Score	20.0 ± 0.54	9.06 ± 0.50	<0.001
Receiving PPI	13% (5)	15% (10)	1.0
Receiving H2RA	23% (9)	16% (11)	0.441
Receiving Any Acid Suppression	33% (13)	28% (19)	0.661
Receiving Erythromycin	10% (4)	4% (3)	0.250
Hypoallergenic Diet	33% (13)	32% (22)	1.0
Receiving Thickened Liquids	39% (15)	39% (27)	1.0
Enteral Tube Present	31% (12)	20% (14)	0.531

Table 2 shows the comparison between those with normal and abnormal Pedi-EAT-10 scores and is notable for higher proportion that underwent VFSS and with abnormal VFSS; higher mean Pedi-EAT-10 scores were seen in subjects receiving thickened liquids and those with enteral tubes. There was no difference in proportion with abnormal Pedi-EAT-10 score and use of reflux therapies, or hypoallergenic formula.

Table 2:

Comparison Between Subjects with Normal Vs Abnormal Pedi-EAT-10 Results

	Pedi-EAT-10 Abnormal (n=32)	Pedi-EAT-10 Normal (n=76)	P-value
Mean Age (months)	8.44 ± 0.78	6.47 ± 0.62	0.052
VFSS Performed	88% (28)	66% (50)	0.033
VFSS Abnormal	96% (27/28)	71% (35/49)	0.007
Pedi-EAT-10 Abnormal	100% (32)	0% (0)
Mean Pedi-EAT-10 Score	19.39 ± 1.11	4.2 ± 0.42	<0.001
I-GERQ-R Abnormal	47% (15)	32% (24)	0.188
Mean I-GERQ-R Score	14.31 ± 1.11	12.46 ± 0.76	0.174
Receiving PPI	16% (5)	13% (10)	0.765
Receiving H2RA	19% (6)	18% (14)	1.0
Receiving Any Acid Suppression	31% (10)	29% (22)	0.821
Receiving Erythromycin	3% (1)	8% (6)	0.671
Hypoallergenic Diet	31% (10)	33% (25)	1.0
Receiving Thickened Liquids	59% (19)	30% (23)	0.009
Enteral Tube Present	44% (14)	16% (12)	0.008

Seventy-one percent of patients had a VFSS performed. Subjects with higher Pedi-EAT-10 were more likely to have a VFSS performed (10.2±0.96 vs 4.78±1.24, p=0.001). Subjects with lower I-GERQ-R were more likely to have VFSS performed (12.05 ± 0.75 vs 15.7±1.01, p=0.005). However, all mean scores for I-GERQ-R and Pedi-EAT-10 were below the established cutoffs, suggesting no concerns for abnormal reflux or feeding difficulties.

20% (22) of patients had an EGD performed; of these 18% (n=4) had microscopic esophagitis, including 1 subject with eosinophilic esophagitis and 1 subject with Helicobacter pylori. EGD findings were not associated with abnormal I-GERQ-R or Pedi-EAT-10 (p>0.616). No patient had an impedance study performed.

Sub-Group Analysis of Subjects with VFSS Results

Of the 77 patients who underwent VFSS, the mean age was 7.1±0.5 months and these patients completed Pedi-EAT-10 and I-GERQ-R at 8.84 ± 0.56 months. VFSS results included 81% (62/77) with abnormal swallow study; 46% (35/77) had aspiration and 35% (27/77) had laryngeal penetration. 91% (32/35) of cases of aspiration were silent.

Among the subjects that had VFSS performed, 35% (27/77) had abnormal I-GERQ-R and 36% (28/77) had abnormal Pedi-EAT-10. Pearson correlation between the scores was 0.369 (p=0.001) for all subjects and 0.351 (p=0.005) for subjects with aspiration/penetration on VFSS, as shown in Figure, Supplemental Digital Content 1. When limiting the analysis only to subjects who completed questionnaires within one month of VFSS, there was still a significant correlation between measures (n=41, r=0.371, p=0.017). Abnormal I-GERQ-R were associated with abnormal Pedi-EAT-10 scores by Fisher’s exact test with RR 2.769 (95% CI 1.044–7.343, p=0.049).

Table 3 shows the comparison between those with normal and abnormal VFSS results and is notable for subjects with abnormal VFSS having younger age, higher proportion with abnormal Pedi-EAT-10 score, higher mean Pedi-EAT-10 score, and lower proportion treated with erythromycin. No significant difference was found for the other comparisons.

Table 3:

Comparison Between Subjects with Normal Vs Abnormal VFSS Results

	VFSS Abnormal (n=62)	VFSS Normal (n=15)	P-value
Mean Age (months)	8.17 ± 0.60	11.59 ± 1.32	0.029
VFSS Performed	100% (62)	100% (15)	1.0
VFSS Abnormal	100% (62)	0% (0)
Pedi-EAT-10 Abnormal	44% (27)	7% (1)	0.007
Mean Pedi-EAT-10 Score	11.34 ± 1.12	5.57 ± 1.10	0.001
I-GERQ-R Abnormal	37% (23)	27% (4)	0.555
Mean I-GERQ-R Score	12.48 ± 0.87	10.27 ± 1.39	0.188

Discriminant Ability of Questionnaire Items

No single I-GERQ-R item was significantly associated with abnormal VFSS results on one-way ANOVA (p>0.086) but scores on question 6 (cry during/after feedings) were higher in subjects with abnormal VFSS (1.10±0.15 vs 0.53±0.22, p=0.04) by t-test.

One of the questions on the I-GERQ-R of particular interest was “does your child have blue spells.” In the cohort overall, 19% (21/108) of subjects answered yes to the blue spell question. This question had relative risk 1.148 (95% CI 1.043–1.264, p=0.142) for predicting aspiration/penetration on VFSS, with specificity 100% and sensitivity 12.9%. All 8 subjects that answered yes to the blue spell question and had VFSS performed had abnormal VFSS results. Because this blue spell question was weighted higher in the score (getting 2 points for a yes question compared to the majority of other questions), answering yes was more likely to result in an abnormal score; of the patients who answered yes to this question, 52% (11/21) had abnormal I-GERQ-R compared to only 32% (28/87) who answered no. In this group of patients answering yes, 10% (2/21) of subjects ended up in the abnormal category because of their affirmative response to this question and these make up 18% (2/11) of those abnormal I-GERQ-R scores.

On the Pedi-EAT-10, questions 3 (swallowing liquids takes extra effort) and 8 (food sticks in throat/child chokes while eating) were significantly associated with abnormal VFSS on one-way ANOVA (p=0.005 and 0.012, respectively). Subjects with abnormal VFSS had higher scores on each of these items compared to those with normal VFSS (1.89±0.19 vs 0.70±0.23 for question 3 and 0.90±0.15 vs 0.13±0.09 for question 8, both p<0.001).

Multiple questions on each questionnaire were associated with abnormal scores on the other questionnaire, as shown in the Table, Supplemental Digital Content 2.

Evaluation of Diagnostic Accuracy of I-GERQ-R and Pedi-EAT-10

On ROC analysis to evaluate the diagnostic accuracy of I-GERQ-R to predict aspiration on VFSS, there was area under the curve (AUC) of 0.640 with p=0.035 for aspiration/penetration on VFSS and AUC 0.592 with p=0.271 for aspiration on VFSS.

On ROC analysis to evaluate the diagnostic accuracy of the Pedi-EAT-10 to predict aspiration on VFSS, there was AUC 0.688 with p=0.024 for aspiration/penetration on VFSS and AUC 0.717 with p=0.001 for aspiration on VFSS, as in Figure 1. Whether a cut-off of 10, 11 or 12 was used on the Pedi-EAT-10, all were associated with abnormal VFSS (p<0.015). The cut-off of 12 had greatest specificity (93.33%) and a cut-off of 10 had greatest sensitivity (54.24%).

Figure 1:

ROC Curve Comparing Pedi-EAT-10 and Aspiration on VFSS

Discussion

When infants and young children are referred to pediatric gastroenterologists for evaluation of feeding difficulties and respiratory symptoms, it can be difficult to discriminate gastroesophageal reflux from aspiration based on presenting symptoms. With both the increasing recognition of oropharyngeal dysphagia and ongoing concerns about acid suppression use in the pediatric population, it is more important than ever to make precise diagnoses (6, 40). In the present study, in order to investigate the overlap between symptoms of reflux and symptoms of aspiration, we evaluated the relationship between validated measures of gastroesophageal reflux and oropharyngeal dysphagia. We found that I-GERQ-R and Pedi-EAT-10 results are highly correlated and both actually have diagnostic accuracy in predicting abnormal swallow study results. The strong overlap between these parent-reported measures supports the idea that symptoms of reflux and oropharyngeal dysphagia might be difficult to differentiate since the symptoms as captured by the questionnaires seem to be indistinguishable.

It is not surprising that there is overlap between the results of these questionnaires based on similar symptom questions but this overlap in their diagnostic capabilities might significantly impact research findings and clinical practice. These instruments have been commonly used in pediatrics and the I-GERQ-R, in particular, has been utilized in numerous studies as a reflux-specific outcome and in many ways has informed our understanding of reflux epidemiology and response to treatment in infants and young children (16–19, 21–23, 26–28). Our results call into question whether the I-GERQ-R is truly measuring reflux alone or if it is actually reflecting oropharyngeal dysphagia. We found that subjects with abnormal I-GERQ-R scores had statistically significantly higher Pedi-EAT-10 scores but subjects with abnormal Pedi-EAT-10 score did not have a statistically significantly higher I-GERQ-R score, suggesting that subjects that would be classified as having abnormal reflux seem to have more swallow dysfunction but not vice versa. This supports the idea that reflux, or at least reflux alone, is not a primary driver of feeding difficulties. Our findings suggest that in cases in which the I-GERQ-R score is abnormal, patients should undergo only a short anti-reflux medication trial and evaluation not just for reflux but also for swallow dysfunction.

We investigated the discriminant ability of each questionnaire’s individual items in comparison to the other score and found associations for several items on each questionnaire, suggesting that there is significant overlap between the clinical syndromes captured in each of these instruments. In the comparison with VFSS results, multiple items from the Pedi-EAT-10 and the I-GERQ-R were associated with abnormal VFSS results. Notably, we found that the blue spell question on the I-GERQ-R in particular might be more associated with aspiration than reflux. While a relatively small proportion of our subjects endorsed blue spells, all subjects with VFSS results that answered yes to this question had abnormal swallow study results and therefore this question had 100% specificity. This is especially noteworthy since this question was given more weight in the I-GERQ-R scoring algorithm and therefore might skew scores toward abnormal for a symptom that has clearly been associated with aspiration and not reflux in as has been shown in multiple prior studies including in infants with severe enough symptoms to be admitted to the hospital for observation(2, 4, 41, 42). These results also raise the possibility that the I-GERQ-R may need modification to reduce the weight placed on this “blue spell” question.

Interestingly, reflux treatments including PPIs, H2 antagonists, erythromycin, and hypoallergenic formula were not associated with lower I-GERQ-R or Pedi-EAT-10 scores in our cohort. Therefore, additional testing with swallow evaluations could be helpful if patients have persistent symptoms despite reflux therapies. While our results show lack of association between taking reflux therapies and I-GERQ-R or EAT-10 scores, more definitive studies will be needed to further investigate this point. We also notably found no significant association between the physician report of reflux symptoms and abnormal I-GERQ-R scores, suggesting that physician and parent impressions of reflux symptoms are not well aligned.

A key limitation of our study is that we only captured a single time point for each subject and therefore can only provide cross-sectional results. Future studies will need to incorporate a longitudinal component to better understand if therapeutic interventions are causally related to changes in scores. Our results are also limited by not all subjects having VFSS results. However, the majority of subjects did have VFSS performed, suggesting our results are likely generalizable. Lastly, our study was performed in a referral population of infants and young children in our Aerodigestive Center, which may have resulted in selection bias such that subjects were more likely to have reflux and/or aspiration and may have had interventions prior to presentation which may have impacted how they responded to the questionnaires. In a general pediatric population it might be that the clinical symptoms of reflux and aspiration have a different degree of overlap and therefore the general practitioner should also maintain an index of suspicion when committing to a diagnosis of reflux without appropriately considering oropharyngeal dysphagia with aspiration.

Conclusions

I-GERQ-R and Pedi-EAT-10 are highly correlated in infants and young children. I-GERQ-R results may actually reflect oropharyngeal dysphagia and not just gastroesophageal reflux disease. Pedi-EAT-10 appears to be a reliable and highly specific predictor of abnormal swallow study results. Investigators and clinicians should be cautious in their interpretation of the clinical significance of I-GERQ-R and consider the significant overlap between symptoms of reflux and oropharyngeal dysphagia in infants.

What is Known / What is New

What is Known

• Feeding difficulties and respiratory symptoms are often attributed to gastroesophageal reflux but may be due to oropharyngeal dysphagia with aspiration

• Infant Gastroesophageal Reflux Questionnaire Revised (I-GERQ-R) is a questionnaire that measures reflux and Pediatric Eating Assessment Tool-10 (Pedi-EAT-10) is a questionnaire that predicts aspiration

• It is not known whether there is overlap between these instruments or if they differentiate patients with aspiration versus reflux

What is New

• I-GERQ-R and Pedi-EAT-10 are highly correlated

• The blue spell question on I-GERQ-R has 100% specificity for predicting abnormal videofluoroscopic swallow study

• I-GERQ-R results may actually reflect oropharyngeal dysphagia and not just gastroesophageal reflux disease in infants