Article Text
Abstract
Objective Gastrostomy facilitates artificial enteral feeding but controversy exists around associated morbidity and mortality. This study aimed to report short and long-term outcomes, and identify parameters associated with overall survival.
Methods A 7-year follow-up audit was undertaken at Aberdeen Royal Infirmary, UK. All patients undergoing endoscopic gastrostomy insertion October 2011–September 2018 were included. Last follow-up was February 2019. Clinical data were prospectively collected. Blood results were retrospectively obtained from electronic records. Statistical analysis was with IBM SPSS V.25.
Results 691 procedures were performed over the 7-year period (520 traditional pull-through percutaneous endoscopic gastrostomy (PEG) and 171 gastropexy procedures to facilitate gastrostomy). Frequency of complications (gastrointestinal bleeding, perforation and peritonitis) was low (each n=1). Overall 7-day and 30-day mortality was 2.2% and 8.4%, respectively. One-year mortality reached 47.6%. There was no difference in survival between PEG and gastropexy procedures (p=0.410). Multivariate analysis reported increased age (p<0.001), increased alkaline phosphatase (p<0.001) and clinical indication (p=0.002) as independently associated with an increased hazard of death. Only age was moderately predictive of mortality (area under the curve 0.74, 95% CI 0.70 to 0.78, p<0.001) in the PEG group. Clinical indication was the only parameter independently associated with mortality in the gastropexy cohort (p=0.003).
Conclusion Endoscopic gastrostomy placement can be safe with a low mortality and low risk of serious complications. Blood markers were not associated with short-term or long-term outcomes. Gastropexy to facilitate gastrostomy is a safe alternative to traditional pull-through PEG procedures. Future work should consider quality of life outcomes to assess the benefit of gastrostomy from a patient perspective.
- PEG
- gastrostomy
- nutrition
Statistics from Altmetric.com
Significance of this study
What is already known on this topic
Gastrostomy facilitates artificial enteral feeding in many clinical conditions.
Morbidity and mortality associated with gastrostomy insertion remain a significant concern.
What this study adds
Endoscopically placed gastrostomy can be safe, with a low mortality and low risk of serious complications.
Gastropexy, to facilitate gastrostomy, is a safe alternative to traditional pull-through endoscopically placed gastrostomy, with similar complication rates and survival data.
Clinical indication for endoscopically placed gastrostomy is independently associated with mortality.
Significance of this study
How might it impact on clinical practice in the foreseeable future
Endoscopically placed gastrostomy should not be discounted due to concern over associated mortality; instead, each case should be assessed individually with consideration of risks versus benefits.
Further endoscopic training in other centres could facilitate wider use of gastropexy techniques when the traditional pull-through percutaneous endoscopic gastrostomy is not possible due to challenging proximal pathology, or to avoid the risk of tumour seeding.
Quality of life outcomes should be a focus of future studies.
Introduction
Gastrostomy facilitates artificial enteral feeding in a vast range of patient groups including neurological disorders, malignant obstruction and those with increased nutritional requirements. First performed in 1979, it is the preferred route for long-term nutritional support when oral intake is insufficient. The National Institute for Health and Care Excellence recommends consideration of gastrostomy in any patient who is malnourished (or who is at risk of malnourishment) and has inadequate or unsafe oral intake with a functional and accessible gastrointestinal tract.1
When a traditional pull-through procedure (hereinafter percutaneous endoscopic gastrostomy (PEG)) is not technically possible, a gastropexy procedure can be considered to facilitate gastrostomy formation (herein gastropexy). Here, a four-point approach is taken in which the stomach is anchored to the abdominal wall with three sutures and a gastrostomy tract is created using a dilator through which a balloon retention device is then inserted. A radiologically inserted gastrostomy (RIG) is a common alternative to an endoscopic approach. Surgical gastrostomies are now rare unless planned as an adjunct to support postoperative recovery.
In 2004, the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) reported concerning morbidity and mortality in patients following endoscopic gastrostomy insertion with an alarming 7-day mortality of 43%.2 Studies from across the world have not supported this report,3–7 however, there is a lack of robust published data from UK centres. Nevertheless, the safety recommendations to (1) assess the potential benefits to the individual, (2) involve the multidisciplinary team (MDT) and (3) develop more comprehensive national guidelines for gastrostomy feeding were sensible and appropriate, although recommended on a modest evidence base.
Patient selection, using risk stratification scoring systems derived from clinical and biological parameters, has been considered8–10 but the results have been variable and scoring systems have not been integrated into routine clinical practice.
We conducted a 7-year follow-up audit at our tertiary gastrostomy referral centre. The primary aim of this study was to report the short-term and long-term mortality following endoscopically placed gastrostomy, with comparison of PEG and gastropexy-facilitated gastrostomy procedures. Additional aims were (i) to report other procedure-related complications and (ii) identify parameters associated with mortality, with subsequent evaluation of whether they could be used to predict outcome.
Methods
Service provision
Aberdeen Royal Infirmary provides a specialist enteral tube feeding service within NHS Grampian, UK. The service has dedicated resources including biweekly protected endoscopic gastrostomy lists, two full-time consultant gastroenterologists, specialist endoscopic gastrostomy nurses (‘abdominal assistants’), clinical nutrition nurse specialists, a defined clinical pathway for patients with motor neuron disease and a large MDT to support inpatient, outpatient and follow-up care. Our standard practice would be to perform a traditional PEG procedure, reserving gastropexy-facilitated gastrostomy for certain clinical indications, that is, obstructing upper gastrointestinal pathology or malignancy to avoid tumour seeding. This decision is made on a case-by-case basis by a consultant gastroenterologist ± radiological input ± multidisciplinary team input. Radiologically and surgically inserted gastrostomy are infrequently performed at our centre.
Study design
All patients undergoing endoscopically placed gastrostomy (PEG or gastropexy) at Aberdeen Royal Infirmary between October 2011–September 2018 were included in the analysis. Last follow-up was February 2019. All patients were entered into the study database by a consultant gastroenterologist who kept this up-to-date throughout the study period. Clinical data prospectively collected included: age; gender; clinical indication for gastrostomy (predefined groups taken from NCEPOD report)2; complications (gastrointestinal bleeding, perforation and peritonitis as defined on CT or requiring surgery); survival (days postprocedure) and mortality (7 days, 30 days, 1 year and overall). Routine haematology (haemoglobin, platelets and neutrophils) and biochemistry (albumin, alkaline phosphatase, sodium, urea and C reactive protein (CRP)) investigations performed prior to the procedure were collected retrospectively from electronic case records in February 2019. Procedures were excluded if no tube was inserted or if it was a re-siting procedure.
Statistical analysis
Analysis was performed using IBM SPSS V.25. Continuous variables are described as median (Q1, Q3). Associations between parameters and gastrostomy procedure (PEG versus gastropexy) were assessed with χ2 test, Fisher’s exact test or Mann-Witney U test. Survival analysis was assessed with Kaplan-Meier (log-rank test) analysis. Multivariate analysis for each subgroup, to investigate the effect of clinical parameters on hazard of death, was by Cox regression. The multivariate model was created using factors significant in Cox regression univariate analysis. Receiver operating characteristic (ROC) analysis was used to report the accuracy of clinical markers thereafter. Hazard tatios (HR) for continuous variables are per 1 unit rise. A two-tailed α level was set at 0.05.
Results
Demographics
A total of 714 patients attended for endoscopically placed gastrostomies between October 2011 and September 2018. Tube insertion was unsuccessful in 29 patients undergoing PEG (7 patients later underwent successful PEG insertion) and 6 patients undergoing gastropexy-facilitated gastrostomy (3 patients later underwent successful PEG insertion and 1 patient later underwent successful gastropexy-facilitated gastrostomy insertion). Of these, 1 patient had unsuccessful gastropexy-facilitated gastrostomy and PEG attempts. Patients who later underwent successful procedures were included within the study period. There were 23 re-siting procedures (9 patients). PEG procedures were performed more frequently (n=520) compared with gastropexy (n=171) procedures (table 1) 36 (28 PEG, 8 gastropexy-facilitated gastrostomy) procedures were performed under general anaesthetic.
The median age of the cohort was 66 (56, 75) years with 382 (55.3%) males and 309 (44.7%) females. The number of serious complications was very low (gastrointestinal bleeding n=1, perforation n=1 and peritonitis n=1). Median values for blood markers for the whole cohort were within the normal ranges for our centre except CRP 17 (4, 41), normal range ≤4. Median time for blood samples to be taken was 1 (1, 1) day pre-procedure. Clinical indications for gastrostomy were diverse and are outlined in online supplementary table 1.
Supplemental material
PEG versus gastropexy
There was no difference in age, gender, number of complications or mortality (7 days, 30 days, 1 year or overall) between PEG and gastropexy groups. Similarly, no differences emerged between the groups when survival against time (Kaplan-Meier, figure 1) was considered (p=0.410). Patients undergoing PEG had higher levels of alkaline phosphatase (p=0.014), urea (p<0.001) and sodium (p<0.001) than those undergoing gastropexy procedures. Venting was only performed as a PEG procedure with the intention to insert a secure, wide calibre tube for drainage.
Survival analysis
Kaplan-Meier survival analysis reported that clinical indication for gastrostomy was associated with overall survival (p<0.001).
Following PEG insertion, the poorest survival was seen in patients with mechanical obstruction to swallowing. The best survival was observed in those with long-term partial failure of intestinal function requiring supplementary intake (figure 2A). Furthermore, within the neurological disease PEG cohort, underlying pathology was associated with survival (p<0.001) (figure 2B).
Following gastropexy-facilitated gastrostomy, patients with oesophageal cancer had poorer survival compared with those with oral/pharyngeal cancer (p<0.001) (figure 2C).
Multivariate analysis reported that increased age (HR=1.037, 95% CI 1.027 to 1.048, p<0.001), increased alkaline phosphatase (HR=1.002, 95% CI 1.001 to 1.003, p<0.001) and clinical indication (p=0.002) were independently associated with an increased hazard of death in the PEG subgroup (table 2). In the gastropexy subgroup, clinical indication for gastrostomy was the only variable independently associated with mortality (p=0.003).
Clinical utility of age and alkaline phosphatase
ROC analysis reported that age was moderately predictive of mortality (area under the curve (AUC)=0.739, 95% CI 0.696 to 0.783, p<0.001), whereas this was not the case for alkaline phosphatase (AUC=0.576, 95% CI 0.522 to 0.630, p=0.007) (online supplementary figure S1).
Discussion
Our overall 7-day mortality of 2.2% is considerably lower than the 43% reported by NCEPOD2 and there are a number of possible reasons for this. Our data report that increased age is independently associated with mortality in patients undergoing PEG procedures and 82% of the NCEPOD cohort (versus 40% in our study) was aged ≥70 years. It could be argued that this may suggest more careful patient selection in our study. However, our cohort had a higher number of patients with malignancy (36% versus11%) which is itself associated with mortality, as demonstrated in certain groups here. The majority of patients requiring venting PEG procedures also had underlying malignancy and the clinical intention was palliation. The differences in mortality are therefore more likely to be related to data selection as the NCEPOD report only included patients where death occurred within 30 days of PEG insertion. Our sample represents ‘real life’ data with a complete dataset of all PEG/gastropexy-facilitated gastrostomy insertions. The NCEPOD report did not demonstrate a reliable denominator for the complication rate and is therefore fundamentally flawed. Our data are in keeping with previously published survival rates from across the world and cover a diverse cohort.3–7
A more recent audit from 2016 generated mortality data similar to ours with 30-day and 1-year mortality at 8% and 35%, respectively,11 although it is important to acknowledge that the main indication for gastrostomy was acute cerebral injury.
We found a very low frequency of procedure-related complications (ie, gastrointestinal bleeding, perforation and peritonitis) in keeping with, but actually lower than, data reported in previous studies.4 These are recognised as serious complications of gastrostomy insertion by the Joint Advisory Group on Gastrointestinal Endoscopy (JAG).12 We did not collect prospective data on less serious stoma site complications (eg, buried bumper syndrome, leakage, ulceration) or aspiration pneumonia but this would be an informative area for future study.
There is very little literature comparing the traditional pull-through PEG with a gastropexy-facilitated procedure. In our practice, gastropexy is reserved for patients where a pull-through PEG is either not possible or technically challenging due to proximal pathology. For example, in patients with a tight, proximally obstructing malignancy it may be unsafe to pull through a gastrostomy tube. In addition, there is a small but definite risk of tumour seeding in the tract or abdominal wall as the gastrostomy tube is pulled through a site of malignancy.13 Thus, for patients undergoing potentially curative cancer treatment, a gastropexy approach is more appropriate. A gastropexy-facilitated gastrostomy itself is more complex and it is reassuring that there was no difference in complications or mortality rates between these groups, especially given the higher prevalence of malignancy in the gastropexy subgroup. These results suggest that gastropexy-facilitated gastrostomy can be a safe alternative to traditional PEG insertion. While patients in the PEG cohort had higher levels of alkaline phosphatase, urea and sodium, this was not considered clinically relevant.
Several previous studies have reported an association between survival and clinical or biological parameters.8–10 In the gastropexy cohort, clinical indication for gastrostomy was the only parameter independently associated with mortality. In the PEG cohort, clinical indication for gastrostomy, age and alkaline phosphatase were all independently associated with mortality. However, when the predictive capability of age and alkaline phosphatase levels were assessed (ROC analysis), only age was moderately predictive of mortality. The discrepancy between statistical significance and clinical relevance is likely explained by entering blood parameters into the multivariate model as continuous variables, done to avoid the use of arbitrary ‘cut-off’ values and to ensure the most robust statistical model was used. No robust evidence was found to support the use of pre-procedure blood biomarkers for predicting mortality following PEG or gastropexy-facilitated gastrostomy. For this reason, we did not attempt to create a scoring system to predict mortality. This approach has previously been explored by other groups but not integrated into clinical practice.9 14 In practical terms, we believe each case should be considered on an individual basis with a balanced MDT assessment of risks versus benefits.
One limitation of our study is that we did not have data to compare endoscopically placed gastrostomy to RIG procedures. The inclusion of RIG patients in this study would have been unfairly biassed, since RIGs are infrequently performed at our centre and generally reserved for patients where endoscopically placed gastrostomy is/has not been possible (eg, due to a large pharyngeal pouch preventing successful intubation). While some studies report one method to be superior over the other, they are often confounded by clinical indication for gastrostomy.15 Given the demonstrated safety of both endoscopic procedures here, the decision regarding endoscopically placed gastrostomy versus RIG may be best assessed by individual centres and based on clinical service provision, staff expertise, patient factors and availability of endoscopy and interventional radiology services.
The use of 30-day mortality to assess the safety of gastrostomy procedures is commonplace,16 17 probably reflecting a traditional ‘end point’ for surgical procedures. Our data demonstrate that both PEG and gastropexy-facilitated gastrostomy have low 30-day mortality rates and a low incidence of serious procedure-related complications. It is therefore likely that 30-day mortality is not reflective of procedure safety itself, but is instead confounded by the underlying clinical indication for gastrostomy. Serious procedure-related complications are most likely to manifest within 7 days, which may be a more sensible time-point to assess procedure-related mortality. However, we also report a very low 7-day mortality rate. While it is important to demonstrate safe practice by monitoring complication rates and mortality, assessing quality of life outcomes in future studies may be more important for assessing the true value of gastrostomy to patients. This could be considered as a useful adjunct to patient selection.
In conclusion, we demonstrate from a large dataset that (1) endoscopically placed gastrostomy can be performed safely to assist with patients requiring long-term artificial enteral nutrition, (2) gastropexy-facilitated gastrostomy can be a safe alternative when a traditional pull-through PEG is not possible and (3) there is no evidence to support the use of blood biomarkers to predict mortality.
Acknowledgments
The authors would like to thank Dr Rebecca McKay (Specialty Doctor at Leeds Teaching Hospital NHS Trust) for her contribution to data collection. The authors would also like to thank Dr Gillian Leggett (Specialty Registrar in Gastroenterology, Aberdeen Royal Infirmary) for contributing to initial study discussions.
Footnotes
Contributors All authors who met the authorship criteria certify they sufficiently contributed to this work. Other contributors are listed under acknowledgements. ELM was the senior/principal investigator for this project. RJP, ELM and AWMcK contributed to study conception and design. ELM collected data prospectively and RJP collected retrospective data. RJP analysed the data and RJP, AWMcK and ELM interpreted data analysis. RJP and ELM wrote the first draft of the manuscript. RJP, ELM and AWMcK contributed to revisions prior to submission.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval Caldicott Guardian approval was granted for this project from Professor Nick Fluck, Caldicott Guardian for NHS Grampian (CG 2018/079).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request.