ARP indicates antireflux procedure; GERD, gastroesophageal reflux disease.
eTable.International Classification of Diseases, Ninth Revision (ICD-9), Codes for Clinical Covariates
McAteer J, Larison C, LaRiviere C, Garrison MM, Goldin AB. Antireflux Procedures for Gastroesophageal Reflux Disease in ChildrenInfluence of Patient Age on Surgical Management. JAMA Surg. 2014;149(1):56-62. doi:10.1001/jamasurg.2013.2685
Gastroesophageal reflux disease (GERD) is a common diagnosis in infants and children, but no objective criteria exist to guide the diagnosis and treatment of this disease in this population. The extent to which age influences decisions about surgical treatment in childhood GERD is unknown.
To identify factors associated with progression to antireflux procedures (ARPs) in children hospitalized with GERD.
Design, Setting, and Participants
Retrospective cohort study using inpatient data from 41 US children’s hospitals in the Pediatric Health Information System database. We included patients younger than 18 years discharged from January 1, 2002, through December 31, 2010, with primary diagnostic codes for GERD (n = 141 190). We evaluated demographics, comorbidities, and diagnostic procedures descriptively and with a multivariate Cox proportional hazards regression model.
Main Outcomes and Measures
Proportional hazard of progression to ARP during admission.
Of the 141 190 patients meeting study criteria, 11 621 (8.2%) underwent ARPs during the study period. More than half of patients undergoing ARPs (52.7%) were 6 months or younger. Although most patients in the ARP group had preoperative upper gastrointestinal tract fluoroscopy (65.0%), these patients did not undergo a uniform workup. The hazard of progression to an ARP was significantly decreased in children aged 7 months to 4 years (hazard ratio, 0.63 [P < .001]) and 5 to 17 years (0.43 [P < .001]) relative to children younger than 2 months. Hazard ratios were independently increased for comorbid diagnoses of failure to thrive, neurodevelopmental delay, cardiopulmonary anomalies, cerebral palsy, and aspiration pneumonia and among patients with tracheoesophageal fistula and diaphragmatic hernia. Each additional GERD-related hospitalization was associated with a 10% increased risk of an ARP.
Conclusions and Revelance
Antireflux procedures are most commonly performed in children during a period of life when regurgitation is normal and physiologic and objective measures of GERD are difficult to interpret. To identify meaningful outcomes after ARP, indications must be clear and standardized. We must clarify the appropriate workup for infants and young children with GERD and better define “failure of medical management” in this population.
Gastroesophageal reflux disease (GERD) is a common pediatric diagnosis that affects as many as 7% of infants and children.1,2 However, clinicians have become increasingly aware of the difficulty in diagnosing GERD and, more importantly, in discriminating GERD from physiologic regurgitation, especially in infants.3 Often, children are diagnosed clinically after showing a response to an initial trial of medications; in other words, the diagnosis is presumed after a successful response of reported symptoms to antireflux medication.2 Childhood GERD is therefore diagnosed most commonly by clinical evaluation, without the use of objective measures. This diagnosis often adheres to patients over time until the symptoms wane as a result of the natural history of regurgitation or an objective test result disproves the presence of the disease.
Although a few reports describe the variation in symptoms and characteristics of this population by age and underlying medical comorbidities, most of the reports are retrospective single-institutional case series, and those reports that involve multiple institutions generally limit their descriptions to the presence or the absence of neurologic disorders.4- 6 Similarly, although some investigations have suggested that infants with GERD are more likely than older children to undergo an antireflux procedure (ARP), to our knowledge no study has examined that trend while controlling for other comorbidities that may serve as indications for ARPs.6
In an effort to understand the population of infants and children with GERD requiring repeated hospitalizations and the subpopulation of this group with progression to ARPs, we identified the population of infants and children hospitalized with GERD to understand the broader characteristics of this group. We hypothesized that infants younger than 2 months would be more likely to undergo ARPs than would older children after adjusting for other clinical covariates.
We conducted a retrospective cohort study of inpatients in the Pediatric Health Information System database, which includes demographic, diagnostic, and treatment data from 41 freestanding children’s hospitals covering 85% of major metropolitan areas in the United States. Discharge diagnoses are assigned using the International Classification of Diseases, Ninth Revision (ICD-9). The study protocol was reviewed and approved by the Seattle Children’s Hospital institutional review board. Informed consent was not required.
We identified all patients younger than 18 years discharged from January 1, 2002, through December 31, 2010, with a diagnosis of GERD (ICD-9 code 530.11 or 530.81). Medical record numbers were used to identify multiple hospitalizations per patient; the patient’s first hospitalization for GERD during the study period was considered the index hospitalization. Patients with only 1 hospitalization who underwent an ARP within 48 hours of admission were excluded because we believed a priori that they represented an inherently different population with a single admission for elective surgical intervention.
Our primary outcome was whether the patient ever underwent an ARP (ICD-9 code 44.66 or 44.67) during the index hospitalization or during subsequent admissions captured during the study period. We examined demographic variables, including sex, age at index admission, and Medicaid status. Our primary exposure of interest was patient age, broken down as younger than 2 months, 2 to 6 months, 7 months to 4 years, and 5 to 17 years, because these ages represent break points in the evolution of physiologic regurgitation in children.7 We also extracted total hospital length of stay (LOS). The following comorbidities were examined based on ICD-9 codes: neurodevelopmental delay, chromosomal anomalies, cardiopulmonary disorders, asthma, Barrett esophagus, esophageal atresia/tracheoesophageal fistula, congenital diaphragmatic hernia, hiatal hernia, abnormality of intestinal fixation (eg, malrotation), cerebral palsy, and seizure disorders (Supplement [eTable]). Comorbid conditions were considered present if the diagnosis appeared during any of the patient’s hospital admissions during the study period; aspiration pneumonia and failure to thrive were measured only during the index admission as a proxy for severity because the aim of the analysis was to identify predictors of progression to ARPs that could be assessed at the initial hospitalization. Although diagnostic studies for GERD are often performed on an outpatient basis for patients undergoing elective procedures, we also examined the relevant diagnostic procedures performed during the hospital stay given that our population had repeated hospitalizations and that all patients received their ARP after at least 2 days of hospitalization. We looked specifically at diagnostic procedures that were performed during the index admission and before the ARP if an ARP was performed.
The patient population was described in terms of demographic and clinical characteristics broken down by ARP status and age group. We used χ2 tests to quantify differences across groups. We used a Cox proportional hazards regression model (with the Breslow method of resolving ties) to identify the hazard of progression to an ARP, controlling for the variables described above. This method accounts for the differing periods of follow-up across patients in whom an ARP might be captured. The survival model was adjusted for clustering at the hospital level to take into account the nonindependence of sampling.
We identified 141 190 patients with a GERD hospitalization meeting study criteria after excluding 3749 patients who had only an elective ARP admission (operation within 48 hours of admission) without a preceding GERD hospitalization (Figure). Comparing this excluded population with our study population confirmed our a priori decision. Whereas 64.0% of the ARP arm of our study population was younger than 1 year with a mean LOS of 36.3 days, 80.8% of the excluded population was 1 year or older with a mean LOS of 5.0 days (data not shown). Of all patients meeting inclusion criteria, 11 621 (8.2%) underwent an ARP during the study period.
Overall, 52.7% of the patients in the study population were aged 6 months or younger. The ARP and non-ARP groups were generally similar with regard to sex and age distribution (Table 1). Medicaid insurance was more common among the ARP patients. Mean total hospital LOS for the index admission was also considerably longer for ARP compared with non-ARP patients (36.3 vs 12.5 days). Compared with non-ARP patients, ARP patients were more likely to present with aspiration pneumonia (11.4% vs 2.4%) and failure to thrive (59.7% vs 28.0%). Each of the comorbidities examined was more prevalent among ARP patients, the most common of which were neurodevelopmental delay (51.7% vs 20.2%), cardiopulmonary disorders (42.0% vs 21.9%), seizure disorder (32.2% vs 11.9%), asthma (21.6% vs 16.5%), and cerebral palsy (19.5% vs 5.7%). Diagnostic procedures were also more commonly performed in the ARP group. Upper gastrointestinal (GI) tract fluoroscopy was the most common diagnostic study performed, with 65.0% of ARP patients undergoing a preoperative study. Other diagnostic studies were administered relatively infrequently by comparison.
Among ARP patients, LOS was significantly longer for younger compared with older patients (70.5 days for patients aged <2 months vs 15.6 days for patients aged 5-17 years) (Table 2). Although failure to thrive was more common among infants, aspiration pneumonia was more common among older children. In general, congenital anomalies (eg, congenital diaphragmatic hernia, esophageal atresia/tracheoesophageal fistula, and malrotation) were more common among infants, whereas neurologic diagnoses (eg, neurodevelopmental delay, cerebral palsy, and seizure disorder) were seen more frequently in older children. Although an upper GI tract endoscopy was more commonly performed in older patients, infants were more likely to receive a preoperative upper GI tract fluoroscopy. Upper GI tract fluoroscopy was the most common study across all age groups. Other diagnostic procedures were performed less commonly and with similar frequency across age groups.
After controlling for all covariates, the proportional hazard of undergoing an ARP was significantly decreased for children aged 7 months to 4 years (hazard ratio [HR], 0.63 [95% CI, 0.54-0.74]) and 5 to 17 years (0.43 [0.36-0.51]) compared with children younger than 2 months (Table 3). Patients 2 to 6 months of age at the index admission had a hazard similar to the youngest patients (HR, 0.96 [95% CI, 0.87-1.06]). As expected a priori, most of the other covariates included in the model were associated with an increased hazard of progression to ARP, with the largest associations noted for hiatal hernia (HR, 4.69 [95% CI, 3.98-5.52]), failure to thrive (2.67 [2.35-3.03]), and neurodevelopmental delay (2.42 [2.17-2.70]). Each consecutive hospitalization for GERD was associated with a 10% increased hazard of ARP (HR, 1.10 [95% CI, 1.08-1.12]), and each consecutive hospitalization with aspiration pneumonia was associated with a 17% increased hazard of ARP (1.17 [1.12-1.22]).
Gastroesophageal reflux disease is a common diagnosis in infants and children, and ARPs remain one of the most common procedure types performed by pediatric general surgeons.8,9 This pathologic entity poses an especially unique challenge in younger children because physiologic regurgitation is common in infancy. Daily regurgitation is noted in about 75% of infants at 2 months of age and 50% at 6 months of age, but this normal reflux resolves spontaneously in most infants by the end of the first year of life.7 Pediatricians and surgeons are charged with the task of determining which cases of regurgitation represent true GERD and which cases of GERD might ultimately warrant surgical therapy. As such, understanding the role of patient age as it relates to current practice in the surgical treatment of GERD is essential in any effort to improve the care of this population of patients. This study is, to our knowledge, the first to examine the independent influence of patient age on the progression to ARP among children hospitalized with GERD. Using a large sample from tertiary pediatric hospitals, our results indicate that among hospitalized children with a diagnosis of GERD, infants younger than 2 months are more likely to receive an ARP than older children are after adjusting for multiple confounding factors.
Ideal management of GERD is a multistep process beginning with an objective diagnosis of pathologic reflux, followed by a trial of validated medical therapy and culminating in an informed decision to offer surgical treatment to appropriate candidates in whom conservative management fails. The degree to which this process occurs in the management of childhood GERD is unclear. Diagnostic criteria in adults are fairly well established because symptoms are generally clear and easily communicated by most patients and because treatment efficacy is easily assessed.10 This approach is more troublesome in children, especially in nonverbal populations (eg, infants and neurologically impaired children). In the past, GERD in children was believed to be adequately diagnosed on the basis of symptoms alone.2 In 2009, however, the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition revised the GERD clinical practice guidelines to be more conservative in their recommendations for diagnosis and treatment.11 Although many authors believe that older children can still receive a diagnosis based on symptoms alone, infants and younger children are generally recommended to undergo objective testing before applying a diagnosis of GERD. Studies have shown that knowledge of diagnostic guidelines by health care providers varies greatly and that infants being treated for GERD often do not meet diagnostic criteria based on objective test results.12,13 The large number of children 6 months and younger in this study in the ARP and non-ARP groups highlights how common this diagnosis is in infants. The number also suggests that physicians may be more likely to apply the diagnosis in this patient group because of diagnostic uncertainty or because other characteristics of these hospitalized infants make it more likely that any regurgitation is perceived as pathologic and indicative of GERD.
Referral for surgical treatment of GERD is generally presumed to be a last resort after failure of medical management, with optimal candidates having undergone specific preoperative evaluations. Indeed, the indications and process are fairly well delineated for adult patients.14 Such guidelines are lacking in children, although several studies15- 17 have shown the utility of esophageal pH studies specifically in identifying children in whom medical management is likely to fail. As evidenced by this and other studies,18- 21 upper GI tract fluoroscopy is frequently used in the preoperative workup among children with GERD. Investigations have shown, however, that upper GI tract studies are poor predictors of pathologic reflux. Regardless of the evidence for or against specific objective studies, surveys indicate that physician practice varies widely regarding the application of these tests, and individual provider gestalt still plays a major role in the decision to offer ARP.22- 24
The results of our study suggest that health care providers are more likely to offer an ARP to infants relative to older children independent of other commonly considered indications for fundoplication in children. The reasons for the difference are unclear but are likely influenced by the high rates of physiologic regurgitation and the diagnostic conundrum in this population, the use of fundoplication as part of a durable feeding plan in infants with failure to thrive, and a greater concern about the potential complications of untreated GERD in younger patients. However, our data show a lack of objective diagnostic studies in all children but especially in infants, a population that certainly warrants greater consideration of confirmatory testing. Similarly, because most cases of infant GERD will resolve with conservative management within 3 to 6 months, our findings of an increased hazard of ARP in the first few months of life suggest that many infants are likely never given an adequate trial of medical management.25 The implications of inappropriate use of ARP in infants are significant, with other studies suggesting that success rates may be lower and recurrence rates higher among these patients.26,27
This study has several limitations. First, the Pediatric Health Information System is an administrative database that is subject to potential miscoding and misclassification. Second, not all potential confounders can be gleaned from this database and controlled for in our model. Third, this analysis focused only on patients with inpatient hospital stays, including those hospitalized specifically for GERD symptoms and those admitted for comorbid conditions. Although not a limitation per se, our choice to focus on a population that has a greater use of inpatient medical resources affects the generalizability of our findings and makes it difficult to apply our findings to older children with isolated GERD in the absence of other significant comorbidities. Last, information on diagnostic studies is limited to inpatient data. In the youngest patients, however, we expect these data should be fairly accurate because many of these patients were likely hospitalized at birth and because the index hospitalization thus captures the period when GERD was diagnosed.
Gastroesophageal reflux disease is a common diagnosis in infants and young children, and the threshold to perform ARPs in these patients appears to be lower. Despite the fact that expert guidelines urge the use of objective studies in the diagnosis of GERD and despite evidence that supports the use of objective studies before performing ARPs, such a standardized evaluation is not common practice. Physiologic regurgitation is common in infants, and even most cases of pathologic reflux respond to conservative measures within the first several months of life. Given what this study shows regarding the current state of practice at tertiary pediatric hospitals, a greater effort is needed to develop and disseminate best-practice standards for the diagnosis and treatment of children, especially infants, with possible GERD. We must clarify the indications for ARP and clarify its use to treat GERD vs its use as an adjunct to a durable long-term feeding plan.
Accepted for Publication: April 5, 2013.
Corresponding Author: Jarod McAteer, MD, MPH, Division of Pediatric General and Thoracic Surgery, Seattle Children’s Hospital, 4800 Sand Point Way NE, Seattle, WA 98145 (firstname.lastname@example.org).
Published Online: November 6, 2013. doi:10.1001/jamasurg.2013.2685.
Author Contributions: Dr McAteer had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: McAteer, Garrison, Goldin.
Acquisition of data: Larison, LaRiviere, Garrison, Goldin.
Analysis and interpretation of data: McAteer, Larison, Garrison, Goldin.
Drafting of the manuscript: McAteer, Larison, Goldin.
Critical revision of the manuscript for important intellectual content: LaRivere, Garrison, Goldin.
Statistical analysis: McAteer, Larison, Goldin.
Administrative, technical, and material support: LaRiviere, Garrison, Goldin.
Study supervision: Goldin.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This paper was presented at the 84th Annual Pacific Coast Surgical Association Meeting; February 17, 2013; Kauai, Hawaii.