Percutaneous Endoscopic Gastrostomy in Children: A Population-Based Study from Iceland,1999–2010

Introduction

In children who are not able to be nourished or take medications by mouth, other routes must be sought. If this is a short temporary condition, intravenous administration or nasogastric tube routes are feasible, but long-term use increases the risk of complications. ¹ Gastrostomy is therefore often required in chronically ill children. The decision to place a gastrostomy can be stressful for parents, ² but studies have shown that most parents are ultimately pleased with the decision. ³ In the past pediatric gastrostomies were performed by open surgery, but now most are performed using endoscopy (percutaneous endoscopic gastrostomy [PEG]), laparoscopy, or a combination of both methods.^4–7 The aim of this study is to examine the indications and outcomes of PEG operations in Iceland.

Patients and Methods

This is a retrospective study that includes all children 18 years of age and younger who received a PEG in Iceland during 1999–2010. Patients were identified by searching records for the ICD surgical codes JDBS10, JDSB00, JDB00, and JDC10. Excluded were those who had an open gastrostomy or a concomitant major open surgery. By chart review we recorded weight and height and calculated Z-scores at surgery and 12 months postoperatively. We also documented the presence of a nasogastric tube prior to PEG placement, diagnosis of patients, complications of PEG placement, and need for fundoplication. Major complications were defined as death, re-operation, admission for intravenous antibiotics, and bleeding requiring transfusion and immediate removal of PEG.

All PEG placements were performed with the pull method. ⁵ In patients deemed unsafe for blind needle insertion and for those undergoing laparoscopic fundoplication, a laparoscopic-assisted PEG was performed with a single supraumbilical incision where a 3.3-mm optic was inserted via the open technique, and CO₂ was insufflated to a maximum of 10 mm Hg in younger children and 12 mm Hg in older children. The gastroscope was introduced into the stomach. The upper abdominal region was inspected with the laparoscope, making sure the space between the abdominal wall and stomach was free. The stomach was then insufflated with the gastroscope, a needle was inserted through the abdominal wall, and the PEG was introduced by the pull method.^5,7 The risk of anesthesia determined by American Society of Anesthesiologists scores ⁸ was obtained from anesthesia records. A Corflo^® PEG kit (Corpak, Buffalo Grove, IL), size 12 or 16 French, was used, depending of the size of the child.

For descriptive statistics we used Microsoft (Redmond, WA) Office Excel™, and Sigmastat (Systat Software, San Jose, CA) was used for statistical analysis. Categorical variables were compared using chi-squared analysis unless sizes were small enough to require use of Fisher's exact test. Continuous variable were compared using Student's t test.

This study was approved by the Icelandic data protection authority and the National Hospital Institutional Review Board committee.

Results

One hundred fourteen gastrostomies were performed during the years 1999–2010. In our study 16 cases were excluded for the following reasons: open gastrostomies (n=10), laparoscopic only (n=3), concomitant open surgery (n=2), and lack of information (n=1). Of the included 98 PEG placements, 85 were by the traditional PEG pull method, and 13 were laparoscopic assisted (10 with fundoplication and 3 because transillumination of the stomach wall failed to provide a safe spot of needle insertion). Median age of the patients was 2 years (range, 1 month–17 years). Fifty-one were girls (52%).

The most frequent underlying diagnosis was neurological disease (Table 1). The majority of patients (87%) were classified as American Society of Anesthesiologists II or III for anesthesia risk (Table 1). In 72% of children a nasogastric tube had been used prior to PEG. Median length of hospital stay was 4 days (range, 1–189 days), with none of the longer admissions due to the PEG placement. Height and weight 12 months after PEG procedure were available for 54 children. Before surgery, median body mass index (BMI) was 14.5 kg/m² (range, 9.8–20.8 kg/m²), and median BMI-for-age z-score was −1.4 (range, −5.9 to 3.0). One year after surgery, median BMI was 15.3 kg/m² (range, 11.2–22.1 kg/m²), and median BMI-for-age z-score was −0.5 (range, −5.1 to 3.8). The median weight increased significantly in 1 year by 1.0 standard deviation (P<.0001; 95% confidence interval, −1.4820 to −0.7387).

One hundred sixty-six complications were recorded in 65 children, of which 96% were minor. The most common complications were granuloma formation (19%) and superficial skin infection (25%). The incidence of infection requiring antibiotics within a week from operation was available in 84 patients receiving PEG insertion: it was not statistically different between those who received preoperative antibiotics versus no antibiotics (5/27 versus 18/57, P=.296). Major complications were 6% and included three children with peritonitis due to gastric leak from the gastrostomy site, one child with an esophageal tear, one child with a buried bumper, and one child with malposition of the gastrostomy tube. Changing the gastrostomy tube to a button at 8 weeks failed in two children with spinal muscular atrophy with intraperitoneal button placement between the abdominal wall and the stomach. We were able to repair this with endoscopy and a guide wire.

Median follow-up time of patients was 47 months (range, 1–152 months). Fourteen children died (range, 1 month to 3 years), but none of the deaths was related to PEG insertion. Twenty-seven children were without gastrostomy at follow-up (Fig. 1), most with primary diagnosis of heart disease or malignancy. Four children developed a persistent gastrocutaneous fistula after removal of a gastrostomy button requiring surgical closure. In 10 children laparoscopic fundoplication was performed at the same time of PEG. Another 12 children (14%) needed Nissen fundoplication 8 months to 6 years later; all of them but one were neurologically impaired.

OPEN IN VIEWER

FIG. 1.

Gastrostomy present at follow-up.

Discussion

In this study we report the outcome of PEG operations over a decade in Iceland. The strength of this study is it includes a whole country where all information and follow-up are in one place. The weaknesses of the study are the relatively small national population size (330,000) and taht it is a retrospective study. The number of PEG procedures performed is comparable to what has been reported in the United States: Gauderer ⁹ estimated that around 11,000 PEG procedures were performed in the United States in the year 2002, which is comparable to our numbers per capita. The rate of major complication in our study seems to be comparable to those in other studies^3,7,10–13 (Table 2). There were no deaths and no perforations of the colon in our population, but the small number of patients might explain this. Two infants needed intervention after failure to change gastrostomy tube to a button at 8 weeks; both of these children suffered from neuromuscular disease with low muscular tone. This raises the question of whether changing the button should be delayed in that population and performed over a guide wire or if a primary laparoscopic gastrostomy should be performed with the button placed directly. The minor complication rate is high but similar to the few reports available elsewhere (Table 2).

In our study we did not routinely administer antibiotics, and there was no statistical difference of infection in those who received antibiotics compared with those did not. This differs from other recent studies such as a prospective study from Italy ¹¹ and a meta-analysis of 1059 patients in 10 studies showing a 15% reduction in the rate of infection in patients receiving antibiotics. ¹⁴

The underlying diseases in our patients were similar in our cohort as reported elsewhere, with the majority of our children being diagnosed with neurological or cardiac disease. The estimated American Society of Anesthesiologists score was II or III in most children, indicating they were ill or young at the time of operation. In over a quarter of patients the gastrostomy was eventually removed because it was no longer needed for enteral access. Similar to other reports this was more likely to occur in children with heart disease and children with malignancy after completion of chemotherapy. ¹⁵ In the majority of the patients, the gastrostomy hole closed spontaneously, but three children required a surgical closure of gastrocutaneous fistula; this value is similar to previously published results.^3,11,16 Fourteen percent of our patients later underwent Nissen fundoplication because of gastroesophageal reflux disease, which is a higher number than in a recent study that reported 9%. ¹⁷ It has been reported that PEG can worsen gastroesophageal reflux disease, ¹⁸ but most infants outgrow their gastroesophageal reflux, ¹⁹ making it difficult to decide on fundoplication at the time of a PEG placement. In our cohort the risk of aspiration was the major factor deciding on choice of operation. The fact that 11 out of the 12 children needing Nissen fundoplication later were neurologically impaired indicates we underestimated the need for fundoplication of this group at the time of PEG placement.

The conclusions we draw from this study are that PEG placement can be performed safely in ill children, that it improves growth, and that it is sometimes temporary. Laparoscopic placement of a button is increasingly the preferred method of long-term enteral access, but there will still be the need for PEG placement where the laparoscopic approach is not available or feasible. Hopefully, these data are helpful in informing parents and physicians.