Laparoscopic versus percutaneous endoscopic gastrostomy placement in children: Results of a systematic review and meta-analysis.

BACKGROUND: Percutaneous endoscopic gastrostomy (PEG) and laparoscopic-assisted gastrostomy (LAG) are widely used in the paediatric population. The aim of this study was to determine which one of the two procedures is the most effective and safe method.
METHODS: This systematic review was conducted according to the preferred reporting items for systematic reviews and meta-analyses statement. Primary outcomes were success rate, efficacy of feeding, quality of life, gastroesophageal reflux and post-operative complications.
RESULTS: Five retrospective studies, comparing 550 PEG to 483 LAG placements in children, were identified after screening 2347 articles. The completion rate was similar for both procedures. PEG was associated with significantly more adjacent bowel injuries (P = 0.047), early tube dislodgements (P = 0.02) and complications that require reintervention under general anaesthesia (P < 0.001). Minor complications were equally frequent after both procedures.
CONCLUSIONS: Because of the lack of well-designed studies, we have to be cautious in making definitive conclusions comparing PEG to LAG. To decide which type of gastrostomy placement is best practice in paediatric patients, randomised controlled trials comparing PEG to LAG are highly warranted.

INTRODUCTION

A gastrostomy tube placement is a frequently performed procedure to benefit children with feeding difficulties. The majority of these children have a significant neurological impairment.[1] Less frequent indications are an inadequate caloric intake in children with chronic medical diseases, failure to thrive, oesophageal atresia, dysphagia, short bowel syndrome and malabsorption.[234]

In 1980, Gauderer et al.[5] introduced percutaneous endoscopic gastrostomy (PEG) as an alternative to the conventional gastrostomy tube placement. Because of its endoscopic approach, this technique is minimally invasive and easy to perform. After the first report, PEG became popular and many authors published on the safety and effectiveness of the procedure.[678910] Nowadays, many institutions have used PEG as standard treatment in children.

Paediatric surgeons, later on, introduced laparoscopic gastrostomy (LAG) as a safe and effective minimally invasive alternative to PEG.[1112] Although both procedures are nowadays widely used, controversy remains on which type is the preferred procedure in the paediatric population.

The main reason for caregivers to eventually choose for a gastrostomy tube placement is an improvement of nutritional status and quality of life (QoL).[13] Improvement of nutritional status and QoL are depended on the efficacy of enteral feeding and post-operative adverse events.[14] Possible adverse events associated with a gastrostomy are gastroesophageal reflux (GER)[15] and complications requiring reinterventions, such as damage to adjacent organs, hypergranulation at the insertion and gastric content leakage at the gastrostomy site.[6]

None of the studies comparing PEG and LAG were able to conclude which type of approach results in the best nutritional outcome and the lowest risk of developing adverse events.[1617]

The aim of this study was to determine which type of approach, either PEG or LAG, would be the best practice in children, through a systematic review and meta-analysis comparing success rates, efficacy on enteral feeding and post-operative adverse events.

METHODS

Search strategy

This systematic review was conducted according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement.[1819] PubMed (1960–2011), EMBASE (1980–2011) and the Cochrane Library (December 2012, issue 6) were systematically searched using predefined search terms to identify all articles comparing PEG to LAG. For PubMed, the following search terms were used: (Paediatrics[Mesh] OR child[MeSH] OR child*[tiab] OR infant*[tiab] OR adolescent*[tiab] OR paediatric*[tiab] OR paediatric*[tiab]) AND (Gastrostoma*[tiab] OR ‘gastric feeding tube’[tiab] OR gastrostomy[MeSH]). The same search strategy was used in EMBASE (replacing ‘[Title/Abstract]’ with ‘ti, ab’ and ‘[MeSH Terms]’ with ‘/exp’). Human, child and adolescent were used as search limits in both databases. Language restrictions and time horizons were not applied.

Study selection criteria

Each article was independently assessed for eligibility using the following predefined criteria:

Study population: Infants and children (0–18 years) who underwent gastrostomy placement

Intervention: Documented surgical technique

Study outcomes: At least one of the outcomes of interest reported below

Study design: Originally published articles.

Studies were excluded from analysis if the authors performed concomitant antireflux procedures or if our primary outcome parameters were not reported. In case of multiple studies reporting on the same study population, only the study with the largest patient population was included.

Outcomes of interest

Primary outcomes of interest were (1) success rates of gastrostomy placement, defined as either completion rate or conversion rate of the procedure, time to first feeding and time to reaching full feedings after operation; (2) efficacy of feeding, defined by the method of feeding and the effects of gastrostomy placement on nutritional status (body mass index, weight-for-height and height-for-age z-scores);[20] (3) QoL, either described or quantified with a validated QoL questionnaire (4) GER, assessed by descriptive symptoms or standardised GER questionnaires and/or 24-h pH monitoring (with or without multichannel intraluminal impedance) and/or the need for post-operative antireflux surgery; and (5) complications associated with the gastrostomy placement (e.g., adjacent bowel injury, early tube dislodgement, intraperitoneal leakage both before and after gastrostomy tube exchange, nonclosure of the gastrostomy after removal of the catheter) and overall rates of complications requiring reintervention under general anaesthesia.

Secondary outcomes of interest were minor complications (e.g., gastric content leakage at the gastrostomy site, stomal infection and hypergranulation of the gastrostomy insertion), operating time, duration of hospital admission and hospital and procedural costs.

Data extraction

Using predefined criteria, titles and abstracts of all retrieved records and subsequently full-text articles were examined for eligibility independently by two authors (Nutnicha Suksamanapun and Femke A. Mauritz). A cross-reference check of included articles was performed to identify articles possibly missed by our search strategy. The following data were extracted separately by the same two authors for all studies meeting the inclusion criteria: Study population characteristics, study design, surgical method, duration of follow-up, number of participating subjects and events for each of the study outcomes of interest. In the case of discrepancies, a third author (Maud YA van Herwaarden-Lindeboom) was consulted to obtain consensus. Methodological quality and risk of bias were assessed for every included study according to the PRISMA-statement[1819] and the Cochrane Collaboration's tool for risk of bias assessment.[21]

Statistical analysis

If more than two studies reported on one of the outcomes of interest, studies were pooled in a meta-analysis. Results were presented as risk ratios (RRs) or weighted mean differences with 95% confidence intervals (CI). The alpha risk was set at 0.05. Data were pooled using the Mantel-Haenszel random-effects meta-analysis model.[22] The random-effects model was chosen to take into account suspected heterogeneity caused by differences in study design and patient population, as it generates a more conservative estimate than analysis using the fixed-effects model.[23] Studies were weighted on sample size and the number of events. Trials with zero events in one arm were included in the analysis by adding a continuity correction of 0.5 to all cells in the two-by-two table for that study. Trials with zero events in both arms were excluded from the meta-analysis. Heterogeneity was calculated using Higgins Chi-square test (P > 0.1).[24] Inconsistency in study effects was quantified by I2 values (I2 > 50%).[2425] Funnel plots were used to help identify the presence of publication bias or other types of bias.[26] All analyses were performed using the Review Manager (RevMan) [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014.

RESULTS

After an extensive literature search, 2347 articles were initially identified. A total of five studies[1617272829] met our in- and ex-clusion criteria [Figure 1]. All studies had a retrospective study design. The five studies reported on a total of 550 PEG and 483 LAG procedures. Gastrostomy placements were performed between 1992 and 2008. Two out of the five studies reported on follow-up time (range 0–135.6 months). All study demographics and surgical techniques are summarised in Table 1, and a risk of bias overview is given in Table 2.

Figure 1

Flow-chart of literature search. PEG: Percutaneous endoscopic gastrostomy, LAG: Laparoscopic gastrostomy

Table 1

Studies (in chronological order) comparing percutaneous endoscopic gastrostomy to laparoscopic gastrostomy in children

Table 2

Risk of bias summary

Primary outcomes

Only two studies reported on the success of the procedure.[1617] The completion rates reported by Zamakhshary et al.[17] were similar for both groups (98% for PEG vs. 100% for LAG). Akay et al.[16] reported on the rate of conversion, which was also similar for both PEG (3,5%) and LAG (4.8%).

Only one study reported data comparing time to enteral feeding.[16] Time to first feeding (0.7 vs. 0.8 days) and time to reaching full feedings (2.1 vs. 2.3 days) were similar for both PEG and LAG.

None of the studies reported data on the efficacy of enteral feeding, nutritional outcomes or QoL after both procedures.

None of the studies reported on GER symptoms or objective GER measurements. Only one study reported that 17 out of 234 patients (7.3%) who received a gastrostomy required an antireflux procedure to treat severe reflux symptoms. This study did not identify a statistically significant difference between PEG and LAG (P = 0.425).[16]

Three studies reported on adjacent bowel injury.[161727] The risk of damaging adjacent intestine was significantly higher during PEG than during LAG (RR = 5.55, P = 0.047); [Figure 2]. There was no adjacent bowel injury in the LAG group. In the PEG group, 2 small bowel perforations and 7 colonic perforations occurred. In one patient, an iatrogenic perforation of the colon was made, and this was identified during the procedure. Consequently, PEG was converted to laparotomy to suture the colonic perforation. All other perforations were discovered postoperatively after the development of peritonitis, sepsis and/or faecal leakage. All patients with perforations after PEG underwent laparotomy and in one patient a colostomy was needed.

Figure 2

Meta-analysis of adjacent bowel injury after percutaneous endoscopic gastrostomy versus laparoscopic gastrostomy. Risk ratios are shown with 95% confidence intervals (Mantel–Haenszel random effects model)

Two studies reported data on early tube dislodgement.[1627] Patients who underwent PEG placement had a higher risk of early tube dislodgement (RR = 7.44, P = 0.02); [Figure 3]. All patients with early tube dislodgement in the PEG group needed a reintervention under general anaesthesia to replace the gastrostomy tube. The only patient in the LAG group with early tube dislodgement received a new gastrostomy catheter as an outpatient procedure without any form of anaesthesia required. The time between initial tube placement and first tube change was mentioned in only two studies and varied between 6 and 8 weeks.[1617]

Figure 3

Meta-analysis of early tube dislodgement after percutaneous endoscopic gastrostomy versus laparoscopic gastrostomy. Risk ratios are shown with 95% confidence intervals (Mantel–Haenszel random effects model)

Only one article reported on intraperitoneal leakage before the first tube change.[17] This study demonstrated similar rates after PEG and LAG (RR = 0.28; P = 0.36). After the first tube exchange, intraperitoneal leakage was reported in two studies.[1617] The relative risk for developing intraperitoneal leakage after tube exchange after PEG was 3.14 compared to those after LAG. Meta-analysis, however, did not identify a statistically significant difference (P = 0.28).

Two studies reported on cases of non-closure of the gastrostomy site after removal of the catheter requiring surgical closure.[1617] The relative risk of non-closure in children who underwent PEG was 0.94 compared to those who underwent LAG. However, meta-analysis did not identify a statistically significant difference (P = 0.92).

Finally, all included articles reported on the number of reinterventions under general anaesthesia. Patients who received a PEG had a relative risk of 2.79 (P = 0.0008) compared to patients who received a LAG (RR = 2.79, P = 0.0008); [Figure 4]. The most frequently reported cause for reintervention in patients with PEG was early tube dislodgement. In patients who underwent LAG the most frequently reported causes were stomal complications (e.g., granulation tissue, erosion, ulceration, non-healing skin and foreign body reaction).[16]

Figure 4

Meta-analysis of all reinterventions requiring general anaesthesia after percutaneous endoscopic gastrostomy versus laparoscopic gastrostomy. Risk ratios are shown with 95% confidence intervals (Mantel–Haenszel random effects model)

To identify possible selection bias, funnel plots were constructed. None of the funnel plots on the primary outcomes showed clear evidence of publication bias, and none of the studies lay outside the 95% CI limits.

Secondary outcomes

Two studies provided data on gastric content leakage to the skin.[2728] Patients after PEG had an RR of 3.82 compared to those after LAG of developing gastric content leakage. However, this difference was not statistically significant (P = 0.15).

Only one article reported on risks of stomal infection requiring treatment.[28] This study did not identify a statistically significant difference (0.4% vs. 0%, RR = 2.87, P = 0.52).

Two studies reported data on operating time.[1617] Akay et al.[16] reported significantly shorter operating time for initial gastrostomy in the PEG group (P = 0.001). Zamakhshary et al.[17] did not identify a difference in operating time [Table 3]. In this study, routine post-operative tube change was included in the overall operating time.[17] These were the only two studies[1617] describing the specific procedure for the PEG, both using the technique described by Gauderer et al.[5] Three studies described the LAG technique.[161729] Techniques described were the ‘four tacking sutures technique’[16] and the ‘two anchoring sutures combined with a purse string suture’.[17]

Table 3

Operating time

Only one study compared procedural costs of both initial procedures.[29] Lee et al. concluded that PEG (1375 USD) resulted in lower procedural costs than LAG (2425 USD), which was explained by the fact that PEG required shorter operating time. However, the costs of the routine second procedure after PEG (to change the gastrostomy catheter to a button) were not included in the total costs.

None of the studies reported exact data comparing hypergranulation, length of hospital stay and hospital costs after both PEG and LAG.

After constructing funnel plots, we did not identify evidence indicating publication bias for these secondary outcomes and none of the studies fell outside the 95% CI limits.

DISCUSSION

In this systematic review and meta-analysis on PEG versus LAG, results show similar success rates in terms of completion or conversion of the procedure, but less adverse events after LAG.

The most important indication for gastrostomy tube placement in children is to provide successful enteral feeding, as this leads to improvement of nutritional status and QoL. Nevertheless, in this systematic review, we found a lack of studies comparing the efficacy of enteral feeding or effects on QoL between PEG and LAG since none of the studies reported on these outcomes.

Two studies reported on the success of the procedure, showing a completion rate of almost 100% after both PEG and LAG[17] and a conversion rate of only 3–5%.[16] This is similar to recently published prospective studies on gastrostomy placements.[63031] The learning curve or level of training of the physician may be of influence on the surgical outcome, however, none of the studies reported details on surgical and/or endoscopic training.

Many publications have investigated the effects of gastrostomy placement on GER.[15323334] However, the exact correlation between gastrostomy and the development of GER remains unclear.[35] In this systematic review, none of the comparative studies between PEG and LAG reported on GER symptoms. Perhaps this lack of published data is caused by the fact that the majority of children requiring long-term enteral tube feeding are neurologically impaired and that specifically in this group evaluation of GER is difficult. Furthermore, GER symptoms are often atypical, and may be disguised by other gastrointestinal problems. Furthermore, normal values of 24-h pH monitoring are not available for children and adolescents, with the exception of early infancy.[363738]

Meta-analysis of serious adverse events identified that patients undergoing PEG had a higher risk (RR 5.55) of injury to the adjacent bowel. PEG is placed from an endoscopic intragastric view, in which a needle is introduced through the abdominal wall without a view on the position of adjacent organs. Major complications, such as adjacent bowel injuries and catheter malplacement are therefore more common after PEG compared to LAG.[39404142] Furthermore, when such complications occurred, endoscopic view could not provide early detection, which may have led to more morbidity. Zamakhshary et al.[17] reported transcolonic tube placement in three children. In all three children, this was diagnosed post-operatively after faecal drainage via the gastrostomy. In none of the patients undergoing LAG adjacent bowel injuries occurred. The laparoscopic approach is possibly safer as it provides a clear intra-abdominal view, thereby preventing adjacent bowel injury.[4344] Moreover, laparoscopic surgery can detect and immediately correct major complications during this primary procedure.

The meta-analysis of early tube dislodgement identified a seven times higher risk of dislodgement for children receiving a PEG than for those receiving an LAG. Tube dislodgement can lead to serious complications, such as intraperitoneal leakage of gastric content, mainly when the gastrostomy was recently performed.[45] During a LAG placement, the stomach is always firmly attached to the abdominal wall with several sutures. If early tube dislodgement occurs, replacement with a new catheter during an outpatient procedure can be performed with a negligible chance of developing adverse events.[16] Patients with a PEG initially only have the gastrostomy tube itself to attach the stomach to the abdominal wall. Therefore, during the 1st week after initial placement, the tract may not be stable enough to safely exchange or reinsert the gastrostomy tube. To underscore this, in this meta-analysis, we found that all patients with early tube dislodgement after PEG required reintervention under general anaesthesia.[16]

A modified T-fastener PEG has been developed to secure a more tight connection and, therefore, it is thought to provide less morbidities in the case of dislodgement.[46] However, some brands of T-fasteners have non-resorbable sutures that need to be removed and sometimes the anchors are not easily removed. Unfortunately, the authors did not separately report on using T-fasteners during PEG. Therefore, comparison of PEG with T-fasteners to PEG without T-fasteners is not possible.

In 8.4% (2.1–19.4%) of children, who underwent PEG a reintervention requiring general anaesthesia was needed, while in children who underwent LAG only 2.5% (0–8.6%) returned to the operating room for reintervention. Meta-analysis confirmed a significant difference (RR = 2.79; P = 0.0008) in favour of LAG. General anaesthesia can be hazardous, especially in patients with cardiac and/or pulmonary anomalies. Therefore, specifically in these patients, LAG should be preferred to prevent multiple (routine) procedures under general anaesthesia.[47]

Minor complications may play an important role in QoL of patients and their caretakers, especially in children with a long-term indication for gastrostomy use.[48] Stomal infection and leakage of gastric contents to the skin were similar after PEG and LAG. Nevertheless, infection rates are difficult to compare because studies do not specify if these infections were confirmed by positive wound cultures or if patients were treated with antibiotics. Hypergranulation is a complication that occurs very frequently after gastrostomy placement. It can cause bleeding and leakage from the gastrostomy site and thus have a major influence on QoL.[6] However, no studies could be identified comparing the incidence of post-operative hypergranulation between PEG and LAG.

PEG is thought to be a faster procedure than LAG; in this study, only two studies compared operating time, of which only one found a significant difference in favour of PEG.[16] In the study by Zamakhshary et al.,[17] PEG and LAG required similar operating time, possibly because the operating time of the second routine procedure (to change the gastrostomy tube to a button) was included in the initial PEG placement time. Even though the initial PEG placement required less operating time, most patients required a second procedure under anaesthesia for routine tube exchange. In patients who underwent LAG, this routine tube exchange could be performed in the outpatient clinic.[1749] The surgeon should take the need for a second procedure under general anaesthesia after PEG into consideration.

Finally, a number of issues still need to be considered regarding the conclusions of this systematic review and meta-analysis. First, all of the included studies in this review were retrospective in design, which resulted in limited access to study outcomes and various forms of bias. Second, patient populations, in terms of comorbidities (e.g., neurologically impaired children, children with congenital cardiac disease and children with cystic fibrosis), were heterogeneous, and studies did not report separate data outcomes comparing specific patient groups. In the meta-analysis, however, none of the outcomes demonstrated excessive heterogeneity. To further take possible effects of heterogeneity between studies into account, we used the random-effects model for meta-analysis, as it generates a more conservative estimate than an analysis using the fixed-effects model.[23] Third, none of the studies used standardised questionnaires or investigation techniques to objectively assess outcomes of efficacy. And finally, patients were operated between 1992 and 2008 and not all authors described their specific surgical techniques in detail. Gastrostomy placement techniques, especially PEG, have changed over the years.[465051] However, the main concept of each procedure (e.g., endoscopic versus laparoscopic view) remained similar.

CONCLUSION

This systematic review and meta-analysis demonstrate that current literature lacks well-designed studies comparing efficacy outcomes, such as QoL, efficacy of feeding and GER and, therefore, we have to be cautious in making definitive conclusions. However, there are retrospective studies available that compare PEG versus LAG and report actual data suitable for systematic review and meta-analysis. These studies cannot simply be discarded and are needed to provide patients, caregivers and referring physicians with evidence-based information on both procedures. Systematic review and meta-analysis of these data show that the success rates in terms of completion of the procedure were similar for PEG and LAG. However, LAG was associated with significantly less serious adverse events, such as adjacent bowel injury and early tube dislodgement, and a lower rate of all reinterventions that require general anaesthesia. Naturally, to make a more informed decision on which procedure of gastrostomy placement is best practice in children, randomised controlled trials comparing LAG to PEG are highly warranted.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.