Pruijsen, J.M.; de Bruin, A.; Sekema, G.; Koetse, H.A.; van Rheenen, P.F.
Objectives: Percutaneous endoscopic gastrostomy (PEG) tube feeding is a convenient method for children requiring long-term enteral nutrition. Preoperative fitness of the majority of pediatric PEG candidates is graded as American Society of Anesthesiologists physical status ≥III, indicating increased risk for peri- and postoperative morbidity. The success rate of endoscopic insertion is high, but variations in the anatomy may lead to failure of PEG placement and repeated exposure to anesthesia for surgical gastrostomy. We evaluated the efficiency of using abdominal plain film with gastric insufflation in the preparatory phase to predict a successful PEG insertion and avoid rescheduling.
Methods: A single-center cohort of candidates for PEG underwent abdominal plain film with gastric insufflation in the preparatory phase before tube insertion. The x-ray film was considered normal when the stomach projected distal to the costal margin. Primary endpoint was the success rate of PEG insertion. Multivariate logistic regression analysis was used to identify factors associated with PEG insertion failure.
Results: A total of 303 candidates for PEG underwent abdominal plain film (age range 0.3–18.1 years). PEG tube insertion succeeded in 287 cases (95%). In case of an abnormal abdominal film, the probability of successful PEG insertion dropped to 67% (95% confidence interval 46%–87%). In a multivariate logistic regression model, significant predictors for PEG insertion failure were spinal deformities (odds ratio [OR] 12.1), previous abdominal surgery (OR 8.5), neurological impairment (OR 4.1), and abnormal plain abdominal film (OR 10.3).
Conclusions: Assessment of the gastric anatomy by abdominal plain film in PEG candidates with spinal deformities, previous abdominal surgery, or neurological impairment may help to identify children with a high likelihood of PEG insertion failure. This strategy enables the endoscopist to notify the surgeon in advance for a potential conversion and avoids repeated exposure to anesthesia.
Percutaneous endoscopic gastrostomy (PEG) is a safe and effective method to provide long-term tube feeding in children (1). An important group of PEG candidates comprises undernourished neurologically impaired children whose preoperative fitness is graded as American Society of Anesthesiologists physical status ≥III, indicating increased risk for peri- and postoperative morbidity (1,2). Repeated exposure to anesthesia is strongly discouraged in this vulnerable group of children. During the endoscopic insertion procedure, the best site for placement of the PEG tube is established by transillumination of light through the abdominal wall and clear visualization of indentation of the stomach by external palpation. In about 5% of PEG candidates, the endoscopic insertion fails (3–7) and conversion to open gastrostomy or a laparoscopic-assisted PEG is necessary (6,8–10). Early identification of patients with a high likelihood of PEG failure could help reducing the number of children exposed to repeat anesthetic sessions. We evaluated the efficiency of using abdominal plain film with gastric insufflation in the preparatory phase to predict a successful PEG insertion and avoid rescheduling.
Study Setting and Participants
We evaluated a cohort of candidates for PEG in the Beatrix Children's Hospital, University Medical Center Groningen, a tertiary care center in the Netherlands. Eligible patients were identified from 4 different hospital registries: surgical procedures registry; hospital admission registry; endoscopy unit registry; and outpatient department registry. By combining these registries, we were confident to have included all of the children that underwent gastrostomy placement between January 2000 and December 2010. Duplicate records were manually deleted and the required information from patient files was entered in a predesigned computerized database. Exclusion criteria were older than 18 years at the time of indication for gastrostomy insertion, gastrostomy placement not performed in University Medical Center Groningen, insertion technique different from the classical pull-through (11), and nonconclusive abdominal plain films.
Abdominal plain films with gastric insufflation were included in the analysis when performed <6 months before gastrostomy placement. The majority of PEG candidates had a nasogastric tube in place at the time of imaging because they were dependent on tube feeding. The others had a tube inserted 1 hour before imaging. In all of the children, the lowest part of the ribcage was marked with a radiopaque chain. With the child placed in supine position, the stomach was insufflated with 50 to 100 mL of air. When the air-filled stomach projected distal to the radiopaque marker, the abdominal plain film was considered normal. When the gastric cavity was only visible behind the rib cage despite air insufflation, the abdominal film was considered abnormal (Fig. 1). Children with an abnormal abdominal film were scheduled for gastrostomy under general anesthesia in the operating room with a surgeon standby in case of any need for conversion to open gastrostomy or a laparoscopic-assisted PEG. In children with a normal abdominal film, the procedure was done without earlier notification of the surgeon.
Data collection was done between March and April 2011 by a single investigator (J.M.P.) and included age, sex, underlying clinical diagnosis, result of abdominal plain film, need for conversion to surgical procedure, and whether this was done during the same anesthetic session. The primary outcome of this study was the success rate of the endoscopic insertion technique. Secondary outcome was the need for a repeated session under general anesthesia. We also aimed to identify predictors of endoscopic failure.
Data were entered on standardized forms and analyzed with SPSS for Windows (version 18.0; SPSS Inc, Chicago, IL). Student t tests and χ2 tests were used to compare baseline characteristics between groups. For nonparametric data, the Mann-Whitney U test was used. All of the tests were 2-tailed. A P value <0.05 was considered significant. Stepwise logistic regression with backward elimination was used to analyze predictors of endoscopic failure. Explanatory factors with a P value <0.10 in univariate analysis were kept in the final multivariate model. Effect size was expressed as odds ratio and 95% confidence interval (CI). The proportion of successful PEG insertions was calculated performing abdominal imaging in the complete study population, as well as after classifying patients as high or low risk for PEG failure with subsequent targeted imaging.
The present study was performed in accordance with the guidelines of the medical ethical committee of the University of Groningen. It involved the study collection of data generated by routine medical care. The data were collected and recorded by the investigators in such a manner that subjects could not be identified, directly or through identifiers linked to the subjects.
A total of 410 children were eligible for inclusion. We excluded 107 patients (26%) on the basis of predefined exclusion criteria (Fig. 2). The characteristics of excluded PEG candidates are shown in Table 1, as well as the characteristics of included children, divided into groups with normal and abnormal imaging results. The groups were comparable in terms of age at PEG insertion and underlying condition. A total of 105 children (35%) had a neuromuscular disorder, with cerebral palsy being the most common condition (n = 40).
Risk Analysis for PEG Insertion Failure
PEG tube insertion succeeded in 287 cases (95%) and failed in 16. Table 2 shows a risk analysis for PEG insertion failure. By univariate analysis, abnormal abdominal plain film, neurological impairment, previous abdominal surgery, and spinal deformities were significant predictors of PEG insertion failure. In the multivariate model, all 4 remained statistically significant predictors. A total of 28 PEG candidates underwent abdominal surgical interventions in the past, of which ventriculoperitoneal shunt placement (n = 11) and bowel resections (n = 7) were the most common indications. Eleven patients had spinal deformities, with scoliosis being the most common abnormality (n = 10).
In 282 of 303 PEG candidates (93%), abdominal imaging showed normal anatomy with the stomach projecting distal to the lowest costal margin. These patients underwent PEG insertion by the pediatric gastroenterologist without earlier notification of the surgeon. In 273 of 282 patients (97%, 95% CI 95–99), the PEG insertion procedure was successful (Table 3). Nine children needed rescheduling on an operating room for surgical gastrostomy placement.
Table 3 shows that 21 of 303 PEG candidates (7%) had abnormal gastric anatomy on abdominal imaging. For these patients, PEG insertion procedure was scheduled in the operating room with a surgeon standby in case of any need for conversion to a surgical procedure. PEG insertion was successful in 14 of 21 cases (67%, 95% CI 46–87), and conversion to surgery was done during the same anesthetic session in 7 children. The downside of this strategy is that 9 other children with normal abdominal imaging had unexpected failure of PEG insertion and were subjected to a second anesthetic session for a surgical gastrostomy.
If a physician would use strategy B—not to notify the surgeon unless at least 1 predictor for PEG failure is present (neurological impairment, previous abdominal surgery, or spinal deformity)—then absence of predictors would give a PEG success rate of 99% (95% CI 98–100). When at least 1 predictor for PEG insertion failure is present, probability of success would be 88% (95% CI 82–94), with conversion to surgery being possible during the same anesthetic session for 14 children. The downside of strategy B is an unnecessary amount of standby time for the surgeon and improper use of expensive operating theatre time (106/303 children, 35% of total number of PEG candidates).
When using strategy C—only PEG candidates with at least 1 predictor for failure are subjected to abdominal imaging—9 patients with PEG insertion failure will be missed (2 low-risk children and 7 high-risk children with normal abdominal imaging). At the same time, the number of patients exposed to radiation is reduced with 60% and the pressure on overstretched operating rooms is eased. The probability of success in the high-risk children group with abnormal abdominal imaging is 53%, and justifies PEG placement in the operating room with the surgeon standing by in case of failure.
The use of abdominal plain film with gastric insufflation in the preparatory phase to assess gastric anatomy predicts success of PEG insertion; however, normal abdominal imaging does not fully exclude PEG insertion failure. In 9 of 303 PEG candidates, insertion failed unexpectedly with insufficient visualization of the indentation of the doctor's finger, most likely caused by a position of the colon or small bowel loop superficial to the distal body of the stomach. In these instances, the endoscopic procedure had to be aborted and the child had to be exposed to a second anesthetic session for surgical gastrostomy. Restricting abdominal imaging to a targeted population with high risk for PEG insertion failure (children with neurological impairments, previous abdominal surgery or spinal deformities) reduces the number of patients exposed to radiation, without increasing the number of nonanticipated insertion failures.
By combining 4 different hospital registries, we were confident to have included all of the children that underwent gastrostomy placement between 2000 and 2010; however, missing data are common in observational research. Routine data sources and clinical databases are often incomplete. We had to restrict our analyses to individuals with complete data on all variables. The characteristics of the individuals with missing data (n = 73) were typical of the whole sample. We also excluded 20 children who were primarily scheduled for surgical gastrostomy. Skipping over the endoscopic procedure was decided by mutual agreement between surgeon and endoscopist in all cases, but may have introduced a selection bias. Eight children with a primarily surgical gastrostomy had predictors of PEG insertion failure (previous abdominal surgery [n = 4], neurological impairment [n = 2], and spinal deformity [n = 2]). In the remaining 12 children, the parents did not give consent for endoscopic placement or gastrostomy placement was combined with another abdominal surgical procedure. In the unlikely event that the imaginary endoscopic insertions would have succeeded in the first 8 cases, the probability of success would have increased to 70% (95% CI 51–89). In the worst case scenario with PEG insertion failures in all 8 patients, the success rate would have dropped to 35% (95% CI 15–55). Inclusion of these patients would have changed the strength of the predictive power of the abdominal film, but not the direction. The radiologists who reviewed the imaging were not blinded for the underlying condition of the child. This may also have caused a bias. Knowing that a PEG candidate belongs to high-risk group may have influenced the assessor to judge the configuration of the stomach bubble more pessimistic.
PEG is the preferred method for enteral access in patients with long-standing insufficient oral intake (12,13). The procedure is safe (1,12), but failure of PEG insertion and consequently rescheduling for a surgical procedure on the operating room is frustrating for the child, the parents, and the physicians (4,14). Assessing gastric anatomy by abdominal plain film before anesthesia helps to identify PEG candidates with a high likelihood of insertion failure and can reduce the amount of repeated exposures to anesthesia. Subjecting only PEG candidates from high-risk groups to abdominal imaging will create a preparatory strategy in which the number of children exposed to radiation balances the limited standby time of the surgeon, without increasing the number of unexpected insertion failures.
We recently sent a short survey to 8 Dutch tertiary care centers for pediatric gastroenterology and 4 large general teaching hospitals with pediatric gastroenterologists. The survey included questions about the preparations before PEG placement. Three centers did not perform abdominal films to evaluate gastric anatomy, 3 centers performed abdominal imaging when abnormal anatomy was expected, and 4 centers subjected all of the PEG candidates to abdominal plain film.
Our “selective strategy” with fluoroscopy for high-risk groups resulted in a PEG insertion failure rate of 3% (9/303 children [95% CI 1–5]). The failure rate in our cohort is in line with a recently described pediatric cohort from the UK with a failure rate of 2% (8/384, 95% CI 1–4). The British study had characteristics similar to ours, although imaging of the abdomen was not performed by default (3). Four patients with cerebral palsy had abnormal gastric position, 2 patients had past abdominal surgery, and the remaining 2 had probable colonic interposition. Abdominal plain film with gastric insufflation could have predicted this outcome in the first 6 children.
We identified 1 adult case series in PubMed that described the use of abdominal plain film 1 day before PEG placement to determine the optimal gastric puncture site (15). A total of 84 patients (of which three quarters with a neurological disorder) had abdominal imaging with insufflation of 500 mL of air through a nasogastric tube. PEG insertion was unsuccessful in 1 patient because the stomach was positioned high behind the ribs, as was already seen on the abdominal film a day earlier. In a second case series, 22 patients with PEG insertion failure caused by inadequate transillumination were referred to the radiology department for fluoroscopy-guided gastrostomy. Sixteen of them turned out to have their stomachs high underneath the lower left rib cage. Choosing the site for fluoroscopy-guided gastrostomy tube placement proved a challenge. Some of the patients even ended up with an intercostal gastrostomy (10).
Our study described an average population of pediatric PEG candidates that is seen in every tertiary care pediatric gastroenterology unit. The characteristics of the participants are comparable with populations described earlier (2,3,16).
In our hospital, children who need long-term tube feeding are primarily presented to the pediatric gastroenterologist for PEG insertion. Although endoscopic gastrostomy is the procedure of choice, other hospitals may decide to primarily offer surgical or radiological methods for gastrostomy placement (8,10,17). Surgical techniques have evidently lower failure rates than the endoscopical techniques, but the procedures are more invasive (6,8).
If, like in our hospital, PEG is the preferred procedure, assessment of the position of the stomach by abdominal plain film in candidates with spinal deformities, previous abdominal surgery, or neurological impairment may help to identify children with a high likelihood of PEG insertion failure. This strategy enables the endoscopist to notify the surgeon in advance for a potential conversion and reduces risk of repeated exposure to anesthesia.
1. Avitsland TL, Kristensen C, Emblem R, et al. Percutaneous endoscopic gastrostomy in children: a safe technique with major symptom relief and high parental satisfaction. J Pediatr Gastroenterol Nutr 2006; 43:624–628.
2. Khattak IU, Kimber C, Kiely EM, et al. Percutaneous endoscopic gastrostomy in paediatric practice: complications and outcome. J Pediatr Surg 1998; 33:67–72.
3. Srinivasan R, Irvine T, Dalzell M. Indications for percutaneous endoscopic gastrostomy and procedure-related outcome. J Pediatr Gastroenterol Nutr 2009; 49:584–588.
4. Kimber C, Beasley S. Limitations of percutaneous endoscopic gastrostomy in facilitating enteral nutrition in children: review of the shortcomings of a new technique. J Paediatr Child Health 1999; 35:427–431.
5. Lowe JB, Page CP, Schwesinger WH, et al. Percutaneous endoscopic gastrostomy tube placement in a surgical training program. Am J Surg 1997; 174:624–628.
6. Pisano G, Calo PG, Tatti A, et al. Surgical gastrostomy when percutaneous endoscopic gastrostomy is not feasible: indications, results and comparison between the two procedures. Chir Ital 2008; 60:261–266.
7. Eger R, Reif S, Yaron A, et al. Percutaneous endoscopic gastrostomy (PEG) in children: indications, the procedure, outcomes, short and long-term complications. Harefuah 2008; 147:21–24.95.
8. Akay B, Capizzani TR, Lee AM, et al. Gastrostomy tube placement in infants and children: is there a preferred technique? J Pediatr Surg 2010; 45:1147–1152.
9. Croshaw RL, Nottingham JM. Laparoscopic-assisted percutaneous endoscopic gastrostomy: its role in providing enteric access when percutaneous endoscopic gastrostomy is not possible. Am Surg 2006; 72:1222–1224.
10. Thornton FJ, Varghese JC, Haslam PJ, et al. Percutaneous gastrostomy in patients who fail or are unsuitable for endoscopic gastrostomy. Cardiovasc Intervent Radiol 2000; 23:279–284.
11. Gauderer MW, Ponsky JL, Izant RJ Jr. Gastrostomy without laparotomy: a percutaneous endoscopic technique. J Pediatr Surg 1980; 15:872–875.
12. Behrens R, Lang T, Muschweck H, et al. Percutaneous endoscopic gastrostomy in children and adolescents. J Pediatr Gastroenterol Nutr 1997; 25:487–491.
13. Gauderer MW. Percutaneous endoscopic gastrostomy and the evolution of contemporary long-term enteral access. Clin Nutr 2002; 21:103–110.
14. Gauderer MW. Percutaneous endoscopic gastrostomy: a 10-year experience with 220 children. J Pediatr Surg 1991; 26:288–294.
15. Chang WK, McClave SA, Yu CY, et al. Positioning a safe gastric puncture point before percutaneous endoscopic gastrostomy. Int J Clin Pract 2007; 61:1121–1125.
16. Beasley SW, Catto-Smith AG, Davidson PM. How to avoid complications during percutaneous endoscopic gastrostomy. J Pediatr Surg 1995; 30:671–673.
17. Akinci D, Ciftci TT, Kaya D, et al. Long-term results of percutaneous radiologic gastrostomy and gastrojejunostomy in children with emphasis on technique: single or double gastropexy? AJR Am J Roentgenol 2010; 195:1231–1237.