Postoperative Surveillance and Detection of Postprandial Hypoglycemia after Fundoplasty in Children

Abstract

Objective

To evaluate the prevalence of postprandial hypoglycemia (PPH) after fundoplasty after the initiation of a universal postoperative glucose surveillance plan in the neonatal intensive care unit (NICU).

Study design

This was a retrospective chart review of children (newborn to 18 years) who underwent fundoplasty at The Children’s Hospital of Philadelphia during the 2-year-period after the launch of a surveillance protocol in the NICU and other units. The rate of screening, frequency of PPH (postprandial blood glucose <60 mg/dL [3.3 mmol/L] on 2 occasions), frequency of postprandial hyperglycemia preceding PPH, timing of PPH presentation, and related symptoms were evaluated.

Results

A total of 285 children were included (n = 64 in the NICU; n = 221 in other units). Of the children screened in all units, 24.0% showed evidence of PPH, compared with 1.3% of unscreened children. Hyperglycemia preceded PPH in 67.7% (21/31) of all screened children. Within the NICU, most children had PPH within 1 week, but only 53.3% exhibited symptoms of dumping syndrome.

Conclusions

This study supports the use of universal postoperative blood glucose surveillance in identifying PPH in children after fundoplasty. Earlier identification of PPH would lead to earlier treatment and minimize the effects of unidentified hypoglycemic events.

Dumping syndrome is a complication of fundoplasty, an operation performed to control severe gastroesophageal reflux disease (GERD). In children, dumping syndrome is characterized by severe postprandial hypoglycemia (PPH) without the significant gastrointestinal symptoms of “early dumping” present in adults.^1,2 PPH, or “late dumping,” occurs 1 to 3 hours after a meal as a result of an exaggerated insulin response, triggered by early hyperglycemia and the potent insulinotropic hormone glucagon-like peptide-1 (GLP-1).^2–4

Although described more than 20 years ago,^5,6 most clinicians remain unaware of the risk of dumping syndrome after fundoplasty. The true prevalence of dumping syndrome in children is controversial; some studies report it as an infrequent complication,^7,8 whereas others report higher estimated frequencies.^9,10 A prospective study estimated that up to 30% of children develop dumping syndrome after fundoplasty, presenting typically in the immediate postoperative period but occasionally years later.¹⁰ However, that study screened only symptomatic patients, many of whom had significant underlying neurologic deficits that might have made symptoms less reliable. Whether the latent presentation of some cases is related to a lack of reliable symptoms, later onset of symptoms, or failure to identify PPH in the immediate postoperative period remains unclear.

Given the unreliability of symptoms, we recommend universal postoperative screening to detect PPH in all children after fundoplasty. In July 2007, we instituted a postoperative surveillance plan in The Children’s Hospital of Philadelphia (CHOP) neonatal intensive care unit (NICU) that allowed monitoring of postprandial blood glucose levels in all children after fundoplasty (see Methods). Earlier identification and treatment of children with PPH should prevent complications of undiagnosed hypoglycemia, such as seizures, developmental delays, and brain damage. We hypothesized that universal surveillance in the immediate postoperative period would improve detection of PPH in children after fundoplasty.

Methods

On July 1, 2007, we introduced a postoperative surveillance plan to the CHOP NICU as standard clinical practice. In children on goal bolus feedings, postprandial blood glucose levels were measured using a bedside glucose meter (OneTouch SureStep; LifeScan, Milpitas, California) at 60, 90, and 120 minutes after feedings for a minimum of 72 hours, and at additional time points if symptoms suggestive of hypoglycemia were identified. A blood glucose measurement <60 mg/dL (3.3 mmol/L) was rechecked immediately with a bedside glucose meter, and if still <60 mg/dL, was confirmed via laboratory analysis of a venous sample. Blood glucose monitoring was discontinued if no level <60 mg/dL was detected within 72 hours after initiation of full feedings. Other units do not routinely perform blood glucose monitoring unless requested by the surgeon. In all cases, critical hypoglycemia (blood glucose <60 mg/dL) was initially treated with dextrose, typically by administration of 2 mL/kg of D10 solution (intravenously or enterally), followed by a recheck of blood glucose level and a change from bolus to continuous feeds.

The surgical database was queried for cases of fundoplasty among all patients admitted to CHOP between July 1, 2007 and June 30, 2009, the 2-year period since the inception of the surveillance protocol. For the purpose of the present study, “fundoplasty” refers to all surgical treatment of GERD, including laparoscopic or open Nissen, Toupet, Dor, and Thal fundoplications¹¹ with or without gastrostomy, but not gastrojejunostomy. A detailed retrospective chart review was undertaken for all patients (newborn to age 18 years) who had undergone fundoplasty. Demographic data and detailed information on the surgical and postoperative courses were collected from the electronic medical records and summarized by unit. To prevent overrepresentation of complex cases, cases were included only if the initial surgical date fell within the study window. Cases were excluded if the patient expired before reaching goal feedings or if a preexisting hypoglycemic condition was present.

The primary outcome was the frequency of PPH in children after fundoplasty after achieving full feedings. PPH was defined as a postprandial blood glucose level <60 mg/dL measured on 2 occasions. Secondary outcome measures included the frequency of glucose monitoring after surgery, frequency of hyperglycemia preceding PPH, time after surgery at which PPH was detected, potential symptoms of dumping syndrome, potential risk factors (eg, gestational age, age, sex, comorbid conditions), and treatment. For the purpose of this study, we categorized hyperglycemia as mild-moderate (blood glucose ≥140 to 179 mg/dL [7.8 to 9.9 mmol/L]) or severe (blood glucose ≥180 mg/dL [10 mmol/L]). Comorbid conditions were subdivided into 4 categories based on discharge diagnoses: neurodevelopmental (eg, seizure disorder), cardiac (eg, congenital heart disease), respiratory (eg, chronic lung disease), and anatomic (eg, gastroschisis).

Statistical Analyses

Data were analyzed using SPSS for Windows version 16.0.1 (SPSS Inc, Chicago, Illinois). Baseline and demographic data were summarized using standard descriptive statistics. The Kolmogorov-Smirnov test was used to assess normality of distribution. Because several variables had skewed distributions, results are presented as median and range, and non-parametric tests were performed whenever applicable. Continuous variables were compared using 2-group t tests or analysis of variance, or their nonparametric equivalents (ie, Mann-Whitney or Kruskal-Wallis test). The χ² or Fisher exact test was used to examine associations between PPH and categorical variables. Statistical significance was defined as a P value <.05.

This study was approved by CHOP’s Institutional Review Board.

Results

Between July 1, 2007 and June 30, 2009, a total of 300 children (newborn to age 18 years) underwent fundoplasty in all hospital units at CHOP. Of these, 285 cases were included in our analysis. Fifteen cases were excluded, due to death before reaching full feedings (n = 10) or repeat surgeries (n = 5) (Figure 1; available at www.jpeds.com).

Figure 1.

Breakdown of subjects after fundoplasty based on hospital unit, screening status, and PPH. Subjects in the NICU underwent postoperative glucose surveillance for the first 72 hours after full feedings were reached, whereas other units did not routinely monitor blood glucose levels unless requested by the surgeon.

Demographic and clinical data are summarized in the Table. Overall, 45.9% of the subjects were premature (<37 weeks gestational age), with 23.9% extremely premature (<32 weeks gestational age). Overall, 42.1% of the subjects had multiple comorbidities, with neurodevelopmental (53.7%) and respiratory (46.7%) conditions most commonly found. The majority of fundoplasties (71.6%) were performed laparoscopically.

Table.

Clinical and demographic data for children after fundoplasty

Locations	n	Age, months, median (range)	Male/female	Gestational age, weeks			Multiple comorbidities (≥2)	Surgical type, open/laparascopy
				<32	32–36	≥37
NICU	64	3 (1–10)	35 (54.7%)/29 (45.3%)	28 (43.8%)	18 (28.1%)	18 (28.1%)	31 (48.4%)	13 (20.3%)/51 (79.7%)
Other units: CICU	38	3.5 (1–69)	22 (57.9%)/16 (42.1%)	3 (9.4%)*	6 (18.8%)	23 (71.9%)	17 (44.7%)	27 (71.1%)/11 (28.9%)
PICU/Surgery/miscellaneous	183	15 (1–271)	100 (54.6%)/83 (45.4%)	31 (19.0%)*	33 (20.2%)	99 (60.7%)	72 (39.3%)	41 (22.4%)/142 (77.6%)
Totals	285	6 (1–271)	157 (55.1%)/128 (44.9%)	62* (23.9%)	57 (22.0%)	140 (54.1%)	120 (42.1%)	81 (28.4%)/204 (71.6%)

PICU, pediatric intensive care unit.

^*26 unknown: (6 in CICU, 9 in PICU, 8 in Surgery, and 3 in miscellaneous units).

Within the NICU, 71.9% of the subjects were premature, with 43.8% extremely premature. Most of these children had comorbid respiratory conditions (67.2%), with 48.4% having more than one comorbid condition. The majority of fundoplasties were performed laparoscopically (79.7%).

Within the cardiac intensive care unit (CICU), a unit devoted to the care of children with severe congenital heart disease, only 28.2% of the subjects were premature, with 9.4% extremely premature. These children had a similar rate of multiple comorbidities (44.7%), but a much higher rate of open fundoplasties (71.1%). Within the rest of the hospital units, 39.2% of the subjects were premature, with 19.0% extremely premature. Of these children in units other than the NICU or CICU, 39.2% had multiple comorbidities, and the majority underwent laparoscopic fundoplasty (77.6%).

PPH

Once the postoperative surveillance plan was introduced to the CHOP NICU, the rate of screening was predictably much higher than in other units, which did not routinely monitor for PPH unless requested by the surgeon. In the NICU, 95.3% of children were screened for hypoglycemia after fundoplasty, compared with 30.8% in other units. As the primary site of screening, the NICU had the highest rate of detected PPH. Regardless of unit, a significant association between PPH and screening was seen (P < .0005).

The primary outcome of this study was the prevalence of PPH in children after fundoplasty since the inception of the surveillance plan. The frequency of PPH was very similar in children screened in the NICU and children screened in other units. In the NICU, 24.6% (15/61) of the children screened had PPH, with 18% (11/61) having a blood glucose level <50 mg/dL. In other units, the frequency of PPH was 23.5% (16/68), with 14.7% (10/68) having a blood glucose level <50 mg/dL. In contrast, PPH was identified in only 1.3% (2/156) of unscreened children. Overall, children with PPH tended to be younger; median age was 3 months (range, 1 to 198 months) in the PPH group and 8 months (range, 1 to 271 months) in the non-PPH group (P < .0005). However, within the NICU, the median age was 3 months in both the PPH and non-PPH groups. Moreover, a significant association was not identified with prematurity (P = .093), surgical type (P = .68), sex (P = .27), or comorbid conditions (all P > .35).

Hyperglycemia often preceded hypoglycemia. In general, children with PPH had a higher frequency of hyperglycemia, both mild-moderate and severe, compared with children without PPH (P = .003). Despite this, the finding of hyperglycemia was not always predictive of PPH. In the NICU, this association achieved statistical significance only when comparing severe hyperglycemia with a combined group including children with mild-moderate hyperglycemia and those without hyperglycemia (P = .025), but not when considering each group individually (P = .055). Although the majority of subjects with PPH in the NICU exhibited postprandial hyperglycemia (66.7% [10/15], including 26.7% mild-moderate [4/15] and 40.0% severe [6/15]), 51.0% (25/49) of those without PPH also had hyperglycemia (38.8% mild-moderate [19/49] and 12.2% severe [6/49]). Among screened children with PPH in all units, 67.7% (21/31) showed the typical pattern of dumping syndrome with early hyperglycemia followed by later hypoglycemia. The prevalence of any glucose abnormality, either hyperglycemia or hypoglycemia, was as high as 69.0% (89/129), but it decreased to 39.5% (51/129) when only severe hyperglycemia was considered. A previous study proposed that a decrease of >108 mg/dL (6 mmol/L) between peak and nadir blood glucose levels be considered diagnostic of dumping syndrome.¹² Approximately 73.3% (11/15) of the patients in the PPH group in the NICU met this criterion, compared with 12.2% (6/49) in the non-PPH group (P < .0005). In other units, 61.1% (11/18) in the PPH group met this criterion, compared with 21.4% (30/140) in the non-PPH group (P = .001).

In our study group, the majority of subjects exhibited PPH within 1 week from the initiation of bolus feedings (66.7% by 3 days; 86.7% by 1 week), and all cases of PPH were identified within 10 days after reaching full feeds (Figure 2). This suggests that the majority of cases of PPH diagnosed during the immediate postoperative period may be detected shortly after full feeds are achieved. Within the NICU, 11 of 15 individuals diagnosed with PPH required continuous feedings or extensive feeding manipulations. Four children in the PPH group resolved before discharge.

Figure 2.

Cumulative frequency of PPH cases in the NICU (n = 15) based on the day PPH was detected after reaching full feedings. The majority of cases were detected by 1 week after reaching full feeds (66.7% by 3 days, 86.7% by 7 days), and all cases were detected within 10 days.

Overall, the presence of symptoms associated with PPH was unpredictable and a somewhat unreliable indicator. Symptoms commonly associated with PPH, including pallor, diaphoresis, irritability, seizures, abdominal distention, and/or diarrhea, are nonspecific. Within the NICU, 35.9% of all subjects showed symptoms consistent with PPH regardless of whether or not they exhibited blood glucose instability. Only 53.0% of the subjects with PPH had concomitant symptoms.

Discussion

Dumping syndrome has long been recognized as a complication of fundoplasty,^5,6 but most practitioners are unaware of this risk, and parents are rarely informed of this potential complication when asked to provide consent for surgery. Our findings suggest that postoperative surveillance improves detection of PPH in children after fundoplasty. PPH was identified at a significantly higher rate in the children screened compared with those in whom routine monitoring was not used and PPH was diagnosed only after symptoms were identified. The unreliability of symptoms and the ubiquitous availability of bedside glucose meters provide a strong argument for universal postoperative screening for PPH after fundoplasty.

Multiple criteria have been suggested as diagnostic of dumping syndrome, but no consensus exists. We limited our definition to PPH, as opposed to any glucose abnormality, for several reasons: first, unlike adults, children exhibit little evidence of early dumping; and second, hypoglycemia may produce neurologic damage and subsequent developmental delays, seizures, and other complications that may be prevented by early detection and correction. Furthermore, the clinical significance of a short period of hyperglycemia remains unclear, particularly in the postsurgical period. Hyperglycemia has also been used to define dumping syndrome,¹⁰ but many children with PPH may go undiagnosed if hyperglycemia alone is used to diagnose dumping syndrome. Although our protocol lacked sensitivity for detecting the hyperglycemic spike often seen in dumping syndrome, whether hyperglycemia is critical to the pathogenesis of dumping syndrome or whether other factors, such as hyper-secretion of GLP-1, are responsible for the exaggerated insulin response to a meal, is unclear.⁴

Our findings suggest that universal surveillance in the immediate postoperative period identifies cases of PPH that otherwise would have been missed had the diagnosis relied solely on the recognition of symptoms. Based on our results, nearly half of at-risk infants might have been missed or not detected until much later without universal screening. This mirrors previously reported findings indicating a significant overlap between symptoms in infants with and without hypoglycemia,¹³ and reaffirms the difficulties of detecting hypoglycemia in infants based on symptoms alone.

Most cases of PPH were detected within the first week after full feeds were achieved. Previous studies have suggested 2 forms of dumping syndrome, with presentation in the immediate postoperative period or much later (mean, 27.6 months after surgery).¹⁰ We propose that the later presenters might have had PPH that went undetected because of a lack of reliable symptoms and failure to screen. Even though the immediate postoperative period represents only a short time period, our plan provides an uncomplicated way to capture PPH in at-risk individuals.

Unfortunately, current treatment options for children with PPH are limited. Different therapies have been used with varying success to treat dumping syndrome, including uncooked cornstarch,¹⁴ pectin,¹⁰ octreotide,¹⁵ acarbose,² and dietary manipulations.¹⁶ PPH persists in many children despite these interventions, and affected children require a regimen of continuous enteral feedings to prevent hypoglycemia but nonetheless remain at high risk for hypoglycemic events if feedings are abruptly stopped. In the NICU, most children with PPH required continuous enteral feedings as sustained treatment. Interestingly, in 4 children with PPH, hypoglycemia had resolved, and bolus feedings were tolerated by the time of discharge. Even though these cases may represent false-positive screens, it remains unclear why PPH is transient in some children and persistent in others.

The present study has several limitations. Foremost, the study is retrospective and limited to records available from clinical care. As with any form of screening, we risk identifying false-positive cases and overestimating the prevalence of PPH. In clinical practice, we recommend performing an oral glucose tolerance test or a mixed meal tolerance test, and consider this the standard of care in cases of suspected dumping syndrome; however, neither confirmatory test was performed in this study. In addition, although our subjects had no obvious clinical evidence of hypoglycemia before the operation, we cannot exclude other possible causes of hypoglycemia, specifically fasting hypoglycemia. Moreover, given the high number of younger subjects with multiple comorbidities, whether our results can be generalized to other populations is unclear. This study also was limited by the short time window available for screening during the immediate postoperative period. Given this limitation, there was no way to capture those who might develop PPH after leaving the hospital, potentially putting a large number of individuals at risk for morbidities related to undetected hypoglycemic events. Specifically, some children remain on continuous feedings at discharge and are not available for inpatient blood glucose monitoring of bolus feeds. Continuous feedings might mask dumping syndrome, and hypoglycemia may emerge only after the reintroduction of bolus feedings. We routinely provide families of children who have undergone fundoplasty with an education sheet on the symptoms of potential dumping syndrome; however, given the unreliability of symptoms, we fully recognize the limitation of this measure. Parents of children who have undergone fundoplasty could be taught to perform home blood glucose monitoring, giving them the opportunity to capture PPH that may present later.

Despite these limitations, our study underscores the need for postoperative blood glucose screening in children after fundoplasty. Many children with PPH would not have been identified had symptoms alone been used to detect cases of suspected PPH. Even when symptoms are present, they are nonspecific and have significant overlap with many other conditions, making identification more difficult without rigorous screening. As such, we recommend screening all patients after fundoplasty and fully disclosing the possibility of this severe complication to all families before surgery.

Glossary

CHOP

The Children’s Hospital of Philadelphia

CICU

Cardiac intensive care unit

GERD

Gastroesophageal reflux disease

GLP-1

Glucagon-like peptide-1

NICU

Neonatal intensive care unit

PPH

Postprandial hypoglycemia