National Trends in the Use of Antireflux Procedures for Children

METHODS

This study was approved by the institutional review board of the University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School. A retrospective cohort study of children (age: <18 years) undergoing antireflux procedures was performed with data obtained from the Healthcare Cost and Utilization Project Nationwide Inpatient Sample from 1996 to 2003. The Nationwide Inpatient Sample is designed to approximate a 20% sample, in each year, of US community hospitals, defined as nonfederal hospitals that are not short-term rehabilitation, long-term care, or psychiatric facilities. Hospitals are identified for sampling on the basis of 5 hospital characteristics, namely, number of beds, teaching status, location (urban versus rural), ownership (public versus private), and US census region. Although children's hospital status is not used as a sampling criterion, children's general hospitals and general hospitals with a children's unit (“hospital in a hospital”) are represented in the databases. Records are provided with a weight that allows calculation of nationwide estimates by accounting for the sampling methods used.

Each record included diagnosis and procedure codes defined with the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9). Antireflux procedures were identified by using diagnosis and procedure codes as described previously.¹⁵ Children (age: <18 years) with an ICD-9 procedure code for esophagogastroplasty (code 44.65) or other procedures for creation of esophagogastric sphincteric competence (code 44.66) were identified. Because ICD-9 code 44.65 was less specific for an antireflux procedure, we excluded records that did not include a diagnosis code for gastroesophageal reflux, esophagitis, esophageal ulcer, or hiatal hernia (code 530.10, 530.11, 530.19, 530.81, 530.20, or 553.3). We also excluded records if a diagnosis code for esophageal cancer (code 150.0–150.9), achalasia (code 530.0), or diaphragmatic hernia with gangrene or obstruction (code 551.3 or 552.3) was present.

Patient-related variables obtained from the database included age, gender, payer source, admission type (elective versus nonelective), hospital teaching status and location (urban versus rural), discharge disposition, length of stay, and hospital charges. Medicaid and Medicare were combined into a single payer source category. Payer sources identified as self-pay, no charge, or other payer types were designated as “other.” Neurologic impairment was identified by modifying categories designated in the Clinical Classifications Software, a tool developed for grouping diagnosis codes into meaningful categories (Appendix).¹⁶ Nine categories of complex chronic conditions were identified by using ICD-9 codes as described previously.¹⁷ Hospital-related variables identified in the database included census region (Northeast, Midwest, South, and West), hospital teaching status (nonteaching versus teaching), and hospital location (urban versus rural).

Analyses were performed by accounting for the survey design of the database (Stata 8.0; Stata Corp, College Station, TX). The population rates of antireflux procedures were calculated by using national estimates and US census data.¹⁸ Dichotomous variables were evaluated for secular trends by using logistic regression modeling year as a continuous variable. A 2-sided Wald test was used for univariate comparisons. Multivariate linear regression analyses were used to determine factors associated with length of stay and multivariate logistic regression analyses were used to identify factors associated with death. Predictors for the multivariate models were selected by using a backward elimination strategy, with the criteria for exclusion set at P > .05. The linear regression model was evaluated by using the adjusted coefficient of determination (adjusted R²). The accuracy of the logistic regression model was evaluated by using area under the receiver operating characteristic curve and the Hosmer-Lemeshow c-statistic.

RESULTS

There were 9987 records of children who underwent an antireflux procedure between 1996 and 2003, yielding a national estimate of 48665 procedures during this period (∼9 procedures per 100000 population). In all years, the highest population-based procedure rate was observed among infants (range: 49–101 procedures per 100000 population) (Table 1). Although procedure rates were generally higher in 2003 than in 1996, substantial year-to-year variations were observed among different age groups and census regions during the study period (Table 1). A significant decrease in the overall population-based rate of antireflux procedures was observed in each year of age until 5 years of age (P < .01), after which no significant differences in procedure rates at each age were observed (Fig 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Rates of antireflux procedures. The bar graph indicates the population-based rates of antireflux surgery for neurologically impaired and normal children. The overall procedure rate decreased in each year of age until 5 years of age (P < .01). The line graph indicates the percentage of children undergoing an antireflux procedure who had neurologic impairment. The percentage of children who were neurologically impaired increased in the first 3 years of life (P < .001).

Compared with children in older age groups, infants were significantly less likely to be neurologically impaired but had more complex chronic conditions. Cardiovascular, respiratory, renal, and gastrointestinal conditions were more common among infants than among older age groups (Table 2). Similarly, the percentage of patients with any chronic condition was highest among infants, decreasing significantly in each older age group (P < .01). The percentages of children undergoing antireflux surgery who were neurologically impaired increased significantly in the first 3 years of life (P < .001) but decreased steadily in the second decade of life (Fig 1). During the study period, there was a significant decrease in the percentages of children undergoing antireflux procedures who were neurologically impaired (Fig 2). When data were analyzed according to age group, this trend was observed among children in each older age group but not among infants (Fig 2). The largest decrease was observed among children 10 to 17 years of age (59% in 1996 and 34% in 2003).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

Percentages of children undergoing an antireflux procedure who had neurologic impairment in each year. A significant declining trend in the percentage of neurologically impaired patients was observed overall (P = .001) and among children 1 to 9 years of age (P = .004) and 10 to 17 years of age (P = .002) but not among infants (P = .13).

The average length of stay was 27 days (Table 2). No significant difference in length of stay was observed between 1996 and 2003 (29 ± 3 vs 28 ± 2 days; P = .77). Independent predictors of longer length of stay among survivors included neonatal admission, 6 of the chronic conditions evaluated, payer sources other than private insurance, and Western hospital location (Table 3). A significant interaction between age and neurologic impairment was observed, with the decline in length of stay associated with increasing age being less for neurologically impaired children. When the interaction term was removed from the model, increasing age was associated with a significantly shorter length of stay and neurologic impairment with a significantly longer length of stay.

Splenectomy and esophageal laceration each occurred in 0.2% of patients. The overall in-hospital mortality rate was 1.3%. Mortality rates were significantly higher during neonatal admissions and among children with neurologic impairment or cardiovascular or respiratory conditions (Table 4). Mortality rates were lower among children treated in the West, compared with those treated in the Northeast or South. The average charge for hospitalizations that included an antireflux procedure was $95 086 (Table 2). A significant decrease in charges associated with hospitalizations for antireflux surgery was observed as age increased (P < .001).

DISCUSSION

The current study provides insight into the use of antireflux surgery since the introduction of laparoscopy for children. Although it has been suggested that the nationwide use of antireflux procedures for children has increased recently,⁷ no identifiable trends in population rates according to age group or census region were observed. This finding contrasts with the dramatic nationwide increase in the use of antireflux surgery among adults that occurred after the introduction of laparoscopy in private hospitals during a similar period.^14,15 It is interesting to note that a similar pattern of use of antireflux surgery among adults was not observed in all hospital types, with procedure rates initially increasing and then decreasing to baseline in Veterans Administration hospitals.¹⁹ Because use patterns in single institutions may differ from national trends,²⁰ our findings do not exclude important local variations in the use of antireflux procedures for children.

Although we did not observe significant changes in the population rate of antireflux procedures, the current study provides evidence that the indications for antireflux surgery have evolved during the laparoscopic era. A decrease in the percentages of children with neurologic impairment undergoing antireflux procedures was observed overall and among children after infancy during the study period. Because administrative databases do not contain sufficient clinical information, we could not identify the specific reasons for these trends. Evolving indications for antireflux surgery for both neurologically impaired and normal children may be contributory. Among neurologically impaired children, potential factors leading to decreased use include a critical reappraisal of the role of “prophylactic” antireflux surgery among those undergoing gastrostomy tube placement,^21,22 improved patient selection resulting from better delineation of symptoms attributable to gastroesophageal reflux,²³ evidence suggesting gastrojejunostomy as a reasonable alternative to an antireflux procedure combined with a gastrostomy tube,²² and the recent availability of more-effective acid-suppression medications (proton pump inhibitors).^5,7 Among neurologically normal children, lowering of the threshold for antireflux surgery related to the perceived benefits of laparoscopic surgery may be contributing to increased use, leading to additional lowering of the percentage of neurologically impaired children undergoing this procedure. More-careful analysis of the specific indications for antireflux surgery among neurologically normal and impaired children will be needed to determine which among these and other proposed factors have contributed to the observed trends in use.

Our results showed that antireflux surgery was performed most commonly for infants (45% of procedures). This finding was consistent with results from a multicenter study from tertiary care centers using data obtained before the laparoscopic era.³ The profile of infants undergoing antireflux surgery differed from that of older children, with chronic conditions being more frequent among infants and neurologic impairment being more frequent among older patients. Surgical series have supported the aggressive use of antireflux procedures for infants with supraesophageal symptoms refractory to medical management, including recurrent aspiration, failure to thrive, and reflux-associated apnea and life-threatening events and have reported generally favorable results in this age group.^24,25 Recent reviews have raised concerns about the extent of use of antireflux procedures for infants because of the transient nature of reflux and the difficulty of attributing symptoms to reflux in infants and have emphasized the importance of careful preoperative evaluation and patient selection for surgery in this age group.^7,26 The predominance of infants among children undergoing antireflux procedures illustrates the potential importance of these concerns.

Although we used administrative data to identify children with neurologic impairment, the overall percentage of neurologically impaired children in the current study was similar to values reported in the prelaparoscopic and laparoscopic eras.^10,12,27,28 We observed that children with neurologic impairment represented a high percentage of those undergoing antireflux procedures, with the highest percentage being observed among children 1 to 9 years of age. When this finding was combined with the age-based procedure rate, the number of neurologically impaired children undergoing surgery was observed to be greatest among infants, declining rapidly over the first 5 years of age. Neurologic impairment was associated independently with longer lengths of stay and higher mortality rates. Identification of neurologic impairment as a risk factor for worse postoperative outcomes is consistent with findings from several previous studies.^8–10

There are important limitations to this study that should be recognized. We did not have access to important clinical data that might have affected or better explained our findings, including the indications for antireflux surgery, procedure-specific complications, and the type of antireflux procedure performed (including the performance of a gastric emptying procedure or revision procedures). Because administrative coding does not accurately identify procedure-specific complications (including gas bloating, dysphagia, and disorders of gastric emptying) that often occur only after discharge, we focused on easily identifiable, in-hospital outcomes, including splenic injury, esophageal injury, length of stay, and death. These outcomes are useful for gaining perspective on the performance of antireflux surgery for children but are not sufficient for complete evaluation of these procedures. Because the Nationwide Inpatient Sample does not contain data regarding the specialty of the operating surgeon and cannot be used reliably to determine surgeon volume, we also could not analyze provider-specific trends in use.

CONCLUSIONS

We observed that the indications for antireflux procedures have evolved during the laparoscopic era. Although population-based procedure rates have not increased, we observed a significant decrease in the proportion of children with neurologic impairment undergoing these procedures. Neurologically impaired children remain at increased risk for death and prolonged length of stay after antireflux surgery. Infants are an important group undergoing antireflux surgery and predominated throughout the study period. The current study highlights the need to evaluate the impact of antireflux surgery among children by using carefully designed, prospective studies.

APPENDIX

The ICD-9 codes for neurologic impairment are as follows (including all levels within each code): anoxic injury, asphyxia: 348.0, 348.1, 768.5 to 768.9, 799.0; chromosomal abnormalities: 758.0 to 758.9; central nervous system (CNS) tumors: 191.0 to 192.1, 225.0 to 225.3; CNS anomalies: 740.0 to 742.9; CNS infections: 323.6 to 323.8, 326, 324.0, 324.9; coma: 348.4, 780.01, 780.03; cerebral palsy: 343.0 to 343.9; developmental delay: 315.31 to 315.39, 315.5 to 315.9; hereditary and degenerative CNS diseases: 330 to 331, 333.4, 333.5, 334.0 to 335.9; intracranial injury with prolonged coma: 850.4, 850.5, 851.0 to 851.9 (with fifth digit 4 or 5), 852.0 to 852.5 (with fifth digit 4 or 5), 853.1 to 853.1 (with fifth digit 4 or 5), 854.0 to 854.1 (with fifth digit 4 or 5); mental retardation: 317 to 319; other nervous system disorders: 325, 332.0, 332.1, 340, 341.0 to 341.9, 348.0, 348.1, 348.30 to 348.9, 358.00, 358.01, 359.0 to 359.3, 779.2; other paralytic disorders: 342.0 to 342.9, 344.00 to 344.9, 438.2 to 438.5; seizure disorders: 345.00 to 345.91, 780.39.