Impact of Fundoplication Versus Gastrojejunal Feeding Tubes on Mortality and in Preventing Aspiration Pneumonia in Young Children With Neurologic Impairment Who Have Gastroesophageal Reflux Disease
OBJECTIVE. Aspiration pneumonia is the most common cause of death in children with neurologic impairment who have gastroesophageal reflux disease. Fundoplications and gastrojejunal feeding tubes are frequently employed to prevent aspiration pneumonia in this population. Which of these approaches is more effective in preventing aspiration pneumonia and/or improving survival is unknown. The objective of this study was to compare outcomes for children with neurologic impairment and gastroesophageal reflux disease after either a first fundoplication or a first gastrojejunal feeding tube.
PATIENTS AND METHODS. This was a retrospective, observational cohort study of children with neurologic impairment who had either a fundoplication or gastrojejunal feeding tube between January 1997 and December 2005 at a tertiary care children's hospital. Main outcome measures were postprocedure aspiration pneumonia–free survival and mortality. Propensity analyses were used to control for bias in treatment assignment and prognostic imbalances.
RESULTS. Of the 366 children with neurologic impairment and gastroesophageal reflux disease, 43 had a first gastrojejunal feeding tube and 323 underwent a first fundoplication. Median length of follow-up was 3.4 years. Children who received a first fundoplication had similar rates of aspiration pneumonia and mortality after the procedure compared with those who had a first gastrojejunal feeding tube, when adjusting for the treatment assignment using propensity scores.
CONCLUSIONS. Aspiration pneumonia and mortality are not uncommon events after either a first fundoplication or a first gastrojejunal feeding tube for the management of gastroesophageal reflux disease in children with neurologic impairment. Neither treatment option is clearly superior in preventing the subsequent aspiration pneumonia or improving overall survival for these children. This complex clinical scenario needs to be studied in a prospective, multicenter, randomized control trial to evaluate definitively whether 1 of these 2 management options is more beneficial.
Aspiration pneumonia (AP) is the most common cause of death in children with severe neurologic impairment.1–4 These children have dysfunctional swallowing and gastroesophageal reflux disease (GERD), which places them at risk for AP.5–7 There are competing surgical and radiologic management approaches to treat GERD and prevent AP in this population.8–10 The surgical option is the antireflux procedure known as the fundoplication. Fundoplication is the third most common procedure performed by pediatric surgeons.11 Approximately 40% of pediatric fundoplications in the United States are in children with neurologic impairment.12 Gastrojejunal feeding tubes (GJTs) offer a nonsurgical alternative for treating GERD in children with neurologic impairment.13,14 Two previous studies compared the efficacy of the 2 therapies and found no difference in preventing subsequent AP.15,16 The cross-sectional design of these important studies hampers clinicians, who still do not know which procedure, a fundoplication or GJT, when performed first, is superior in reducing subsequent AP or improving survival for these children.
Although a multicenter, randomized, control trial (RCT) would be the best approach to answering this question, it seems premature to undertake 1 for 2 reasons: (1) single centers have small numbers of eligible patients; and (2) there is strong personal preference and institutional culture that guides clinicians' choice between these procedures. In short, there is an absence of compelling data for clinical equipoise across institutions.17 We were able to use a large administrative database containing clinical, laboratory, and radiologic data to compare outcomes of the 2 procedures and to use propensity scores to partially overcome the problem of confounding by indication. We created an inception cohort of children with neurologic impairment and GERD and used entry criteria as if they were being enrolled into a RCT. The objective of our study was to compare the time to develop AP and survival in children with neurologic impairment who were treated for GERD with either a first fundoplication or a first GJT. If our study failed to show a clear superiority of 1 procedure over the other, ie, demonstrated clinical equipoise, this would provide strong justification for and interest in a future multicenter RCT.
The study was conducted at Primary Children's Medical Center (PCMC), a 233-bed tertiary care children's hospital that serves as both the primary hospital for Salt Lake County and as the tertiary referral hospital for 5 states (MT, ID, WY, NV, and UT).18 PCMC is owned and operated by Intermountain Healthcare Inc, a large, not-for-profit, vertically integrated health care system with 20 hospitals and 50 outpatient clinics.
Patients were identified using Intermountain Healthcare's Enterprise Data Warehouse, an organized, integrated, and searchable administrative database that stores 8 million patient encounters and includes clinical, laboratory, and radiologic data from all inpatient and outpatient settings and uses a linked unique identifier for each individual patient.19 The Enterprise Data Warehouse is used with more frequency for data after January 1, 1997.
We enrolled patients meeting the following criteria: (1) date of birth between January 1, 1997, and December 31, 2005; (2) having a previously published International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code for neurologic impairment,12,20 with the following additions, demyelinating central nervous system conditions (340–341.9), other paralytic conditions (344.0–344.9), spina bifida (741.0–741.93), spinocerebellar disease (334.0, 334.9), and tuberous sclerosis (759.5) either on the date of the procedure or in any previous encounter with Intermountain Health care; (3) diagnosis of GERD (defined by ICD-9-CM codes 530.11 or 530.81), either on the date of the procedure or in any previous encounter; and (4) having either a first fundoplication (44.66, 44.67) performed or a first GJT (internal charge code) placed between January 1, 1997, and December 31, 2005, at PCMC (see Fig 1). GJTs were image-guided feeding tube placements after a previously placed gastrostomy feeding tube. Patients with neurologic impairment who had GERD but neither study procedure (and only medical management with acid suppression or prokinetic agents) were excluded from the main analyses. Rates of AP and death in this group were obtained, although a priori we believed that this represented a different group of children. In addition, patients who were born before January 1, 1997, were excluded to follow the cohort over the study period, and assign them correctly to the appropriate treatment group. Patients were assigned to a treatment group based on the first procedure they received. The study was approved by the institutional review board of the University of Utah Health Sciences Center.
Baseline variables recorded included demographic data, presence of comorbid conditions, previous AP (507.0), presence of a tracheostomy (v44.0, v55.0), chronic lung disease (496, 770.7), seizures (345.0, 345.1, 345.9, 780.39), and the presence of a cerebrospinal fluid shunt (v45.2). We noted the diagnosis of any of the following conditions that may be relative surgical contraindications: microgastria (750.7), Barrett's esophagus (530.85), peptic stricture (530.3, 750.3), esophageal dysmotility (530.5), and short esophagus (530.6, 750.4). Baseline was defined as the date of either procedure (fundoplication or GJT). Comorbid conditions were defined using Complex Chronic Conditions (CCCs), which use ICD-9-CM codes to classify children into 1 of 9 groups of medical conditions (by systems) that are expected to last at least 12 months and involve at least 1 organ system severe enough to require specialty pediatric care and hospitalization at a tertiary care center.21 CCCs have been used to predict mortality in previous studies.22–25 We excluded the category for neuromuscular ICD-9-CM codes because all of the children had neurologic impairment.
The primary outcome measures for this study were the time from either procedure to the next admission to PCMC for AP, defined by the ICD-9-CM code 507.0 and time to mortality after either procedure using the Utah State Vital Statistics Record database. Both events were followed for each patient until October 31, 2006. Descriptive data on the cause of death were obtained by chart review.
Differences between patients undergoing a fundoplication and those undergoing a GJT were compared by using χ2 or Fisher's exact tests for categorical variables; the 2-tailed, unpaired t test, or Wilcoxon rank-sum test was used for continuous variables. The association of fundoplication with mortality was determined by using bivariate and multivariable Cox proportional hazards regression analyses. These Cox models were adjusted for heterogeneity, defined as baseline variables that were associated with mortality and clinically significant variables. The proportional hazards assumption was confirmed by inspection of survival curves and examination of time-dependent covariates. Survival curves were constructed using Kaplan-Meier estimates with comparisons between curves based on the log rank statistic. Similar analyses were performed for the outcome, time to next admission for AP.
Because patients with neurologic impairment and GERD were not randomly assigned to GJT or fundoplication surgery, potential confounding by indication was adjusted for by developing a propensity score for fundoplication surgery treatment.26,27 Stepwise logistic regression analyses were used to select baseline variables that were associated with fundoplication. Variables that were clinically relevant but not significant in the initial logistic regression analyses were then added to derive a full nonparsimonious model. This model yielded a concordance index (c) index of 0.78, indicating a strong ability to differentiate between patients undergoing a GJT versus a fundoplication. Using these selected baseline variables, a propensity score for undergoing a fundoplication was estimated by maximum likelihood logistic regression analysis. This score ranged from 0.23 to 0.98 and reflected the probability that a patient would undergo a fundoplication. Additional Cox proportional hazards analyses were adjusted for confounding by using propensity scores as a variable in the models.
Appropriate regression diagnostics including examining residuals, testing for outliers, and multicollinearity were performed to confirm the validity of these analyses. All analyses were performed by using SAS 9.1 (SAS Institute, Inc, Cary, NC).
Of the 366 children with neurologic impairment and GERD, 43 had a first GJT placement and 323 children underwent a first fundoplication during the study period. Median length of follow-up until death or October 2006 was 3.4 years. Table 1 provides a comparison of the baseline characteristics of the children. Those who received a first fundoplication were younger (P < .001); less likely to have a cerebrospinal fluid shunt (P < .001); less likely to have a tracheostomy (P < .001); more likely to have a cardiovascular condition (P = .014); less likely to have a gastrointestinal (P = .002), congenital or genetic defect (P = .028), or malignancy (P = .04); and less likely to have a relative surgical contraindication (microgastria, Barrett's esophagus, peptic stricture, esophageal dysmotility, or short esophagus; P < .001).
Association of Baseline Features With Mortality and AP
Unadjusted analyses were performed to assess the relationships between baseline features and mortality or AP for the total cohort. Children who had spinocerebellar disease (P = .006), demyelinating central nervous system disorders (P = .03), and cerebral degeneration (P = .001) were more likely to die. Children who had brain or spinal cord anomalies (P = .02) and cerebral palsy (P = .001) were associated with having AP. Children with a tracheostomy (P = .06) and chronic lung disease (P = .07) had a trend toward being associated with AP.
Fundoplication Versus GJT and Associations With Mortality and AP
In the total cohort (n = 366), 49 (13%) patients died and 55 (15%) developed AP (see Appendix⇓ for the causes of death). There was no difference in survival or time to AP between the 2 groups in the unadjusted Cox proportional hazards analyses (see Figs 2 and 3). For death, the unadjusted hazard ratio (HR) for those who received a first fundoplication as compared with a first GJT was not significant (HR: 0.54; 95% confidence intervals [CIs]: 0.26, 1.10; P = .09). Heterogeneity adjusted models also revealed nonsignificance (HR: 0.55; 95% CI: 0.25, 1.21; P = .14). Variables used in the propensity score were those that reached significance, age (P = .01), cerebral spinal fluid shunt (P = .006), tracheostomy (P < .001), and relative surgical contraindications (P < .001). Propensity adjusted models revealed a trend toward increased survival for those patients who received a fundoplication (HR: 0.49; 95% CI: 0.23–1.03, P = .06). When the cohort was stratified by age in the Cox model that included propensity scores, patients older than 1 year had a lower mortality (HR: 0.30; 95% CI: 0.12–0.73; P = .008). For AP, none of the models revealed significance. The result for the propensity adjusted model was HR: 0.71; 95% CI: 0.21–1.69; P = .44.
In this largest-to-date cohort study of children with neurologic impairment and GERD, whom we were able to follow from birth, we found the time to develop AP did not differ between those treated with an initial fundoplication and those treated with an initial GJT. There was no difference in mortality associated with first fundoplication compared with first GJT.
Our findings are similar to the study by Wales et al,15 who performed a retrospective observational study comparing GJT to fundoplication in a similar population of children. The proportion of children with a subsequent AP was similar to our findings: 15% for the fundoplication group and 23% for those who were treated with GJT, ie, no significant difference between groups. The investigators noted that over the 10-year period of their study, there had been a cultural shift in their institution. In the first portion of the study period, children were more likely to receive a fundoplication first, whereas the opposite was true in the latter half of the study. Confounding by indication was not addressed in their study. Goldin et al28 examined reflux-related hospitalizations after antireflux operations for all children in Washington. They reported that there was no reduction in AP after antireflux operations in children with developmental delay. Neither of these studies restricted patients to be able to follow them from birth throughout their lifetime.
Our study addresses the clinical question faced by clinicians and parents at the initial point that a child with neurologic impairment and GERD is thought to require an intervention, ie, which of the 2 options is the most likely to prevent subsequent AP and increase survival? We designed this study to capture patients at the time that they could have been randomized to their first procedure. We used propensity scores as previous studies have demonstrated that if inherent biases such as nonrandom treatment assignment are controlled for, observational studies can achieve similar estimates of the effects of therapeutic RCTs.29,30 By restricting our patients to an inception cohort, we could identify which procedure was performed first and then determine the natural history of outcomes of both treatments. Our data demonstrate that rates of AP and time to AP are equally present in either procedure.
In our study, the time to death was similar between the 2 treatment groups, but the study was underpowered to detect true differences in this infrequent outcome. The proportion of patients who died as reported by Wales et al ranged from 12.5% to 17.5% and is similar to our study. The study by Albanese et al16 had lower mortality rates but also more loss to follow-up and shorter follow-up of their patients. Our study included deaths recorded in state vital statistics records, and included all children who died in Utah. In addition, our cohort of patients was younger and may provide a more accurate lifetime assessment of time to death for children with neurologic impairment and GERD. For the limited detailed data we have on death, it seems that complications from either procedure as a cause of death are rare.
Our study has several limitations. Patient assignment was not random; parents and physicians were allowed to choose which treatment they wanted for the child's GERD. We used propensity scores in an attempt to overcome potential confounding by indication.26,27,31,32 However, this method is limited when there are unmeasured variables that may influence the choice between either a GJT or a fundoplication. The only way to completely overcome confounding by indication would be to conduct an RCT where unmeasured variables would be equally distributed among the 2 treatment groups. The unbalanced numbers in the 2 groups is another limitation.
We limited left truncation by restricting our analyses to patients born after January 1997, but this may have restricted the generalizability of our results from this single center study to older children with neurologic impairment who receive surgical or radiologic treatment for GERD. Clinically, we have observed that the majority of these procedures are first performed in younger children (unless the neurologic impairment is acquired, eg, traumatic brain injury in older children). Patients may have been admitted to another hospital, and thus missed in the outcome of AP in our study; however, the children's hospital serves as the tertiary care referral center for the Intermountain West.
The diagnoses of GERD (for our inclusion criteria) and AP (our primary outcome) were based on ICD-9-CM codes. To validate our ICD-9-CM codes for GERD, we examined all 902 fundoplications performed in any child over the study period. In 89% of those cases, the ICD-9-CM codes for GERD were present before the procedure, giving us some assurance that these patients had GERD. Clinically, GERD and AP are difficult diagnoses to make, but we have no reason to suspect systematic bias in diagnosing either between treatment arms. AP may have continued to be caused by primary aspiration (eg, of secretions) and not secondary aspiration (eg, refluxed gastric contents). Only a limited group of children had a swallow study in their evaluation before a first procedure to diagnose primary aspiration; also, it is possible that a child with neurologic impairment may develop primary aspiration depending on the cause of their neurologic impairment. Given the retrospective nature of the study, we could not distinguish between primary and secondary aspiration.
We were unable to determine whether either treatment group continued with oral intake. If continued oral intake were more common in one of the groups it could contribute to differences in the frequency of pneumonia not attributable to the procedure. Detailed data on mortality were available for <60% of patients (given how many died out of hospital). No functional limitations adjustment could be performed because of lack of information in the database. We did not measure other important outcomes, such as quality of life of either caregiver or child over time compared with baseline measures,20 nor complications or cost of either treatment arm, as these were beyond the scope of our study. GJT alone could impact quality of life given (1) the frequent replacements needed in the radiology suite of an experienced institution and (2) the potential added burden of continuous feeds to the child and family.33
We recognize that our study does not address the important question of when should 1 offer either of these procedures to these children? For 671 children (who did not receive either procedure), the AP rate was 5.7% and 8.8% died, over the study time frame. We believe that the lower AP rate reveals that this was most likely a different group of children, who were believed to be less likely to develop AP and thus were not offered either procedure. We have no data regarding whether these procedures were or were not offered to these children. Answering the questions of why these children had similar death rates and the reasons for their deaths will require additional studies.
Despite these limitations, our findings and those of other observational studies indicating a lack of evidence to support one approach over the other provide strong evidence for clinical equipoise. This finding is a necessary precursor to justify a multicenter RCT to determine which therapy is superior. More research is needed for the population of children with neurologic impairment who share several common problems (eg, involving airway and gastrointestinal tract), especially given that the proportion of this population in inpatient settings in the United States has increased by 46% from 1997 to 2003 (Kid's Inpatient Database study, unpublished findings, Berry and Srivastava). Under entry conditions similar to a RCT, our study confirms the previous literature that has been unable to demonstrate a difference between therapeutic options in preventing AP or improving survival.15,28 Only an adequately powered, multicenter RCT that includes an accurate assessment and comparison of child and caregiver quality of life, complications, cost, morbidity and mortality between these 2 therapeutic options can determine if 1 approach is truly superior. Results from such a trial would provide a sound basis for improved counseling of families of neurologically impaired children regarding which treatment option to choose.
Children with neurologic impairment and GERD who are treated with a first fundoplication compared with a first GJT experience similar times to develop AP and mortality after either procedure, when adjusting for bias in treatment assignment by using propensity scores. These therapies should be compared in a future multicenter RCT.
Dr Srivastava is the recipient of a National Institute of Child Health and Human Development career development award K23 HD052553, and this project was supported in part by the Children's Health Research Center at the University of Utah and Primary Children's Medical Center Foundation.
We are indebted to Laurie Coburn, Tanner Coleman, Flory Nkoy, Matt Swenson, and Nick Tsu for their help on obtaining data for this study, and to the anonymous reviewers of the manuscript.
- Accepted May 1, 2008.
- Address correspondence to Rajendu Srivastava, MD, FRCP(C), MPH, University of Utah Health Sciences Center, Division of Inpatient Medicine, 100 North Medical Dr, Salt Lake City, UT 84113. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Children with neurologic impairment commonly suffer from GERD and AP. AP is the most common cause of death in these children. Fundoplication and GJT are comparative treatments that have been shown to have similar rates of AP and death in cross-sectional studies, following either procedure.
What This Study Adds
Propensity scores were used to adjust for confounding by indication and demonstrated no difference in time to readmission for AP or death. The study cohort comprised young patients to account for first procedure performed, and time-to-event rates were provided for future trial planning.
- Maudsley G, Hutton JL, Pharoah PO. Cause of death in cerebral palsy: a descriptive study. Arch Dis Child.1999;81 (5):390– 394
- Spitz L, Roth K, Kiely EM, Brereton RJ, Drake DP, Milla PJ. Operation for gastro-oesophageal reflux associated with severe mental retardation. Arch Dis Child.1993;68 (3):347– 351
- ↵Fonkalsrud EW, Ashcraft KW, Coran AG, et al. Surgical treatment of gastroesophageal reflux in children: a combined hospital study of 7467 patients. Pediatrics.1998;101 (3 pt 1):419– 422
- ↵Lasser MS, Liao JG, Burd RS. National trends in the use of antireflux procedures for children. Pediatrics.2006;118 (5):1828– 1835
- ↵Godbole P, Margabanthu G, Crabbe DC, et al. Limitations and uses of gastrojejunal feeding tubes. Arch Dis Child.2002;86 (2):134– 137
- ↵Ampofo K, Gesteland PH, Bender J, et al. Epidemiology, complications, and cost of hospitalization in children with laboratory-confirmed influenza infection. Pediatrics.2006;118 (6):2409– 2417
- ↵Feudtner C, Christakis DA, Zimmerman FJ, Muldoon JH, Neff JM, Koepsell TD. Characteristics of deaths occurring in children's hospitals: implications for supportive care services. Pediatrics.2002;109 (5):887– 893
- Feudtner C, Hays RM, Haynes G, Geyer JR, Neff JM, Koepsell TD. Deaths attributed to pediatric complex chronic conditions: national trends and implications for supportive care services. Pediatrics.2001;107 (6). Available at: www.pediatrics.org/cgi/content/full/107/6/e99
- Feudtner C, Silveira MJ, Christakis DA. Where do children with complex chronic conditions die? Patterns in Washington State, 1980–1998. Pediatrics.2002;109 (4):656– 660
- ↵Feudtner C, Silveira MJ, Shabbout M, Hoskins RE. Distance from home when death occurs: a population-based study of Washington State, 1989–2002. Pediatrics.2006;117 (5). Available at: www.pediatrics.org/cgi/content/full/117/5/e932
- ↵Pasta DJ. The Lewin Group. Using propensity scores to adjust for group differences. Available at: www2.sas.com/proceedings/sugi25/25/st/25p261.pdf. Accessed November 21, 2008
- ↵Goldin AB, Sawin R, Seidel KD, Flum DR. Do antireflux operations decrease the rate of reflux-related hospitalizations in children? Pediatrics.2006;118 (6):2326– 2333
- ↵Friedman JN, Ahmed S, Connolly B, Chait P, Mahant S. Complications associated with image-guided gastrostomy and gastrojejunostomy tubes in children. Pediatrics.2004;114 (2):458– 461
- Copyright © 2009 by the American Academy of Pediatrics