Predictors of failure of percutaneous endoscopic gastrostomy tube placement: a retrospective study in a tertiary care center in the USA

Article information

Clin Endosc. 2024;.ce.2024.118

Publication date (electronic) : 2024 October 16

doi : https://doi.org/10.5946/ce.2024.118

Ryan Xin^,1, Cassandra Sanossian², Melissa Fazzari³, Brandon Mui⁴, Marouf Hossain⁴, Jennifer Katz⁵

¹Department of Gastroenterology, Montefiore Medical Center, Bronx, NY, USA

²Division of Gastroenterology, Donald and Barbara Zucker School of Medicine at Hofstra/Northwell, Hempstead, NY, USA

³Department of Biostatistics, Montefiore Medical Center, Bronx, NY, USA

⁴Department of Medicine, Icahn School of Medicine at Mount Sinai Hospital, New York, NY, USA

⁵Department of Medicine, NYU Langone Health, New York, NY, USA

Correspondence: Ryan Xin Department of Gastroenterology, Montefiore Medical Center, 111 E 210th St, Bronx, NY 10467, USA E-mail: rxin@mail.einstein.yu.edu

Received 2024 May 9; Revised 2024 September 9; Accepted 2024 September 10.

Abstract

INTRODUCTION

Percutaneous endoscopic gastrostomy (PEG) placement is a common procedure that is used for enteral feeding when a patient is unable or unwilling to meet nutritional needs by mouth. Since its inception in 1980, it has become the gold standard for long-term access of enteral nutrition. It is estimated that over 150,000–200,000 PEGs are performed annually in the United States of America.1,2 It is a well-tolerated procedure, with success rates usually ranging from 90% to 95%.3-6 The most common indications include dysphagia secondary to stroke, dementia, or malignancy, or the need for mechanical ventilation/tracheostomy.

A multitude of studies have examined PEG complications and mortality or examined success rates of PEGs in certain populations. However, in our search, there were no studies specifically examining predictors of PEG success or failure. While the success rate for PEG placement is quite high, the high volume of PEG placements leads to many aborted procedures. Furthermore, the increasing utilization of other techniques such as radiologically inserted gastrostomy tubes has unveiled a need for better stratification to predict which patients are more likely to have an unsuccessful PEG placement.7

PEG placement begins when the flexible endoscope is inserted into the mouth and advanced through the esophagus and into the stomach. After examination of the esophageal and upper gastrointestinal (GI) tract mucosa is done, an external site is found by examining the patient's abdomen to see if the light from the endoscope inside the patient's stomach is transmitted to the skin, a maneuver termed transillumination. Additionally, the endoscopist can press on the abdominal skin in the area of transillumination to see if the indentation from the endoscopist’s finger is well-demarcated along the anterior stomach wall; this maneuver is termed “one-to-one”. The procedure is typically aborted when there is inadequate transillumination or poor “one-to-one”, both subjective decisions made by the endoscopist. Given these anatomical restraints, it was hypothesized that higher body mass index (BMI) might negatively impact PEG success rate given the increased difficulty of both transillumination and one-to-one in these patients. In addition to BMI, this study also analyzed a multitude of other patient, endoscopist, and case characteristics aiming to identify risk factors to prognosticate when PEG placement is likely to be aborted.

METHODS

A retrospective chart review was conducted of patients >18 years old in the Montefiore Health System who were scheduled to undergo inpatient PEG placement from 2016 to 2020 (n=1,138). Inpatient PEG requests were first reviewed by fellows and attendings on the gastroenterology consult service to determine patient candidacy for placement. Each patient’s anatomy was reviewed on imaging, if available, prior to PEG placement and poor anatomic candidates were advised to pursue alternative methods of enteral nutrition access. Failed PEG placement was defined as failure in any portion of the procedure including inability to complete the endoscopic examination, aborting PEG placement because of endoscopic findings, or failure to place the trocar. Demographic and procedure data including patient BMI, indications for PEG, type of anesthesia, location of procedure, and endoscopist characteristics were assessed. BMI was categorized into four classes: underweight (<18.5 kg/m²), normal (18.5–24.9 kg/m²), overweight (25–29.5 kg/m²), obese I (30–34.9 kg/m²), and obese II (≥35 kg/m²). Indications for PEG were categorized into dementia, dysphagia secondary to malignancy, dysphagia secondary to stroke, need for ventilator, or other. Types of anesthesia included monitored anesthesia care (MAC) involving an anesthesiologist titrating propofol or a similar anesthetic, moderate sedation involving use of sedatives such as fentanyl and midazolam, general anesthesia involving an anesthesiologist titrating anesthesia requiring intubation, or no sedation. Location of procedure was classified into endoscopy suite, intensive care unit (ICU), or hospital floor. Endoscopist characteristics included type of attending (service attendings, who performed procedures with fellows, and private attendings, who performed procedures without fellows). Endoscopy reports in ProVation Medical software were analyzed for successful placement of PEGs; pertinent endoscopic findings as well as reasons for unsuccessful attempts were noted.

Patient characteristics and endoscopist characteristics were summarized using descriptive statistics, both overall and stratified by whether the PEG was successfully placed. Univariate associations between characteristics and PEG placement were tested using t-tests, chi-square or Fisher exact test, as appropriate. A multivariable logistic regression model for the probability of successful PEG placement was estimated, which included statistically significant characteristics in univariate analysis or those of a priori clinical interest (BMI, indication). Two-sided p-values less than 0.05 were considered statistically significant.

RESULTS

Patient characteristics

Baseline patient characteristics including age, gender, BMI, and indications for PEG were obtained and stratified by successful PEG placement (Table 1). Eighty-nine percent (1,017/1,138) of PEGs were successfully placed. The mean patient age was 71 years. Seventy six percent of patients had hypertension, 41% had diabetes, 8% were on dialysis, 34% were on a ventilator. The mean BMI was 25.7 kg/m²; 20% of the patient population had a BMI that was underweight, 31% were normal weight, 27% were overweight, and 21% were obese class I or II. The indications for PEG placement were as follows: 31% of PEGs were placed due to dysphagia secondary to stroke, 27% for dysphagia secondary to dementia, 24% for ventilator use, 6% for dysphagia secondary to malignancy, and 12% for miscellaneous causes. Miscellaneous indications included persistent encephalopathy, seizures, neuromuscular issues, primary esophageal dysmotility, and malnutrition. Neither indication (Table 1; p=0.06) nor BMI (Table 1; p=0.16) was significantly associated with successful PEG placement. Patient comorbidities such as hypertension, diabetes, cirrhosis, dialysis, use of anticoagulation, use of mechanical ventilation, or previous aspiration were also examined. Being on anticoagulation was defined as warfarin exposure within 5 days, low molecular weight heparin within 24 hours, or apixaban/rivaroxaban within 1 to 3 days (based on creatinine clearance). None of these comorbidities were significantly associated with differences in successful placement.

Table 1.

Baseline characteristics overall, and stratified by successful PEG placement

Procedural characteristics

Seventy-two percent of procedures were done in the endoscopy suite compared with 16% on the hospital floors and 12% in the ICU. PEG placement success was significantly associated with location of placement (Table 1; p=0.02). PEGs placed in the endoscopy suite or hospital floor had a success rate of 88% and those placed in the ICU had a success rate of 96% (Table 1; p<0.001). Notably, private attendings performed a higher proportion of cases in the ICU than in other locations. While private attendings accounted for 35% of the PEGs placed overall, they placed 56% of the total PEGs placed in the ICU. For anesthesia during cases, 68% of cases were done using MAC, 30% of cases were done using moderate sedation, and 2% were done using general anesthesia. There were only 22 total cases that utilized general anesthesia. Anesthesia type was significantly associated with PEG placement (Table 1; p<0.001). MAC had a success rate of 89%, moderate sedation had a rate of 92.5%, and general anesthesia had a success rate of 54.5%.

Table 2 shows the results from the multivariable logistic regression mode for successful PEG placement. Private attendings were significantly more likely to successfully place PEGs compared to service providers (odds ratio [OR], 7.4; p<0.001) and use of MAC anesthesia was associated with 2.5 times higher odds of success compared with non-MAC anesthesia (p=0.004). Whereas patients with BMI classified class II obesity were observed to have lower odds of successful PEG placement compared with normal BMI patients, the association of BMI and PEG placement was not statistically significant (OR, 0.54; p=0.09).

Table 2.

Multivariable model for successful PEG placement

DISCUSSION

We found that PEG placement is an overall successful procedure with a success rate of 89%, similar to previously reported success rates of 90% to 95% in prospective studies.4,8 Our success rate may be lower due to the inclusion of solely inpatient PEG placements. The mean age of our population was 71 years, indicating that PEGs were largely placed in an elderly population which aligns with the most common indications of stroke and dementia. The mean BMI in our cohort was 25.7 kg/m² and about 50% of the population was overweight or obese, reflecting the demographics of our patient population.

Our findings indicate that, in our cohort, the type of attending (service vs. private), location of PEG placement, and anesthesia type all had significant effects on successful PEG placement. Notably, BMI and indication of PEG placement were not significantly associated with PEG success rate.

While the primary aim of this paper was to find patient characteristics that would predict success or failure of PEG placement, we found that endoscopist and procedure characteristics, rather than patient characteristics, had a significant impact on the success rate of the procedure. None of the studied patient comorbidities (BMI, indication, hypertension, diabetes, anticoagulation, being on mechanical ventilation, cirrhosis, dialysis, previous aspiration event) were significantly associated with differences in PEG success rate.

We had initially hypothesized that BMI may impact PEG success because of the importance of anatomical features during successful PEG placement. Because approximately half of the PEG procedures failed because of poor 1:1 or transillumination which could be exacerbated by increased adipose tissue, we expected BMI to be a simple way to stratify a patient’s risk for unsuccessful PEG placement. When examining previous studies in PEGs, we found that both underweight BMI and overweight BMI were listed as risk factors for worse outcomes after PEG placement.8-10 Surprisingly, BMI did not have a significant effect on successful PEG placement in our study, perhaps indicating that patient abdominal anatomy plays a less important role in the success of PEG placement than previously thought. However, it is important to note that each patient was first reviewed by the gastroenterology consult service to determine whether the patient was an appropriate candidate for PEG placement. This initial selection process may confound the importance of patient anatomy and BMI in the included patients, as patients who were poor candidates for PEG placement based on BMI or abdominal anatomy may have been filtered out during the initial gastroenterology consult.

In addition to BMI, we also examined the indication for PEG as a possible predictor of PEG success. Given that some PEGs were placed due to mechanical reasons with the GI tract such as an obstructive malignancy, whereas others were placed for practical reasons such as the need for long-term mechanical ventilation, we thought that different indications might influence success of the procedure, especially when considering patients’ anatomic features. We found that the indication for PEG placement did not have a significant effect on success rate. However, in our study, there was a much smaller proportion of patients that underwent PEGs for dysphagia secondary to malignancy compared to prior studies.8,11-13 At our institution, these PEGs are typically placed by interventional radiology. Comparatively, there was a relatively high percentage (24%) of PEGs placed for the need for long-term ventilation; only one other study even listed long-term ventilation as a major category of indications.11 Our study was conducted solely of inpatients at a tertiary care center in New York City with a high ICU capacity and included several months during the coronavirus disease 2019 pandemic, increasing the number of mechanically ventilated patients. When examining recent data, we found that there has been a steady increase in the overall number of patients requiring long-term ventilation and a concurrent increase in the number of long-term acute care centers able to accommodate them.14 In 2008, the annual increase in the percentage of patients requiring prolonged mechanical ventilation was five times the increase in overall hospital admissions.15 Taken together, these findings may represent a shift in common indications for PEG placement.

In contrast to patient characteristics, our study showed that endoscopist characteristics had a more significant effect on PEG placement success. We found that there was a significant difference in PEG placement success rate depending on whether the attending was a service or private attending. Service attendings, who perform cases with gastroenterology fellows, were significantly less likely to successfully place PEGs compared with private attendings. The success rate for service attendings was 85.3% in comparison with 97% for private attendings. Multiple explanations exist to explain this phenomenon. A large portion of the failed placements were due to poor 1:1/transillumination, a subjective decision made by the endoscopist which may be determined by their level of risk aversion. Fellows are often the primary proceduralists and have significantly less endoscopy experience than that of attendings. The service attendings may thus, in general, be more cautious because the fellow acts as the primary proceduralist. Furthermore, service attendings have an obligation to teach fellows and may strive for “perfect” one-to-one and transillumination to ensure fellows have appropriate examples from which to learn.

Specific characteristics of each endoscopic case were also analyzed. Location of PEG placement and type of anesthesia given were significantly associated with successful PEG placement. The highest rate of PEG placement success was 96% in the ICU followed by 88% for the floor and endoscopy suite. However, this finding was likely confounded because private attendings were more likely to place PEGs in the ICU, making up 35% of all PEG cases but 56% of ICU cases. Private attendings were significantly more likely overall to place PEGs successfully compared with service providers. The overall high success rate of PEG placement in the ICU could also be confounded by the severity of patient condition and the associated increased scrutiny placed upon PEG placement. In our facility, PEGs are an elective procedure and require an initial evaluation by a gastroenterologist in order to be placed. Patients who undergo PEG placement in the ICU would undergo additional scrutiny given their critical status to make sure they are stable enough for PEG placement. Moreover, these patients are usually more sedated given the multiple indications for artificial sedation in the ICU, increasing the likelihood of successful placement due to less patient interference.

When examining types of anesthesia, MAC and moderate sedation were both similar in success rate at 88% and 93%, respectively, but patients undergoing general anesthesia had a much lower success rate at 50%. Patients undergoing PEG placement in the ICU underwent moderate sedation given the lack of anesthesiology personnel there. Given the increased success rate of PEGs placed in the ICU, this may be a confounder to the increased success rate of moderate sedation compared to MAC. It is also important to note that relatively few patients received general anesthesia, but this group had a large effect size given the severely reduced success rate. Given the effect of general anesthesia on patient hemodynamics, this finding was not unexpected. In addition, patients undergoing general anesthesia are usually sicker at baseline. They may have other comorbidities that make PEG placement more difficult, such as ileus that can cause bowel overlying the PEG tract.

Overall, the main strength of this study is its large sample size and the database of clinically relevant patient, endoscopist, and procedural characteristics, which allowed us to explore many potential predictors of successful PEG placement. To our knowledge, factors predicting PEG success or failure has not been previously studied. There were also several weaknesses to the study. This study was a retrospective single-center study of inpatient PEG placements at a large, tertiary-care institution, limiting the generalizability of the study. As these were inpatient PEG placements, each patient was already seen by a gastroenterology consult team which act as a preliminary screening tool whose practices differ by institution. The findings may be indicative of local practices rather than national standards and may not be applicable to outpatient PEG placement. In addition, as a retrospective chart review, this study was heavily reliant on accurate data input. While providers often strive for accuracy, there sometimes can be confusion over small details such as type of anesthesia (MAC vs. moderate sedation) or ambiguity involving reasons for failed PEG placement. Lack of data regarding type of PEG placement (push vs. pull) further limits generalizability.

In conclusion, the results of our study confirm previous data indicating that PEG placement generally is a successful procedure. This study adds data to show that providers are unable to use patient characteristics to predict when placement of inpatient PEGs might fail. In our cohort, dysphagia secondary to ventilator use comprised a larger proportion of PEG placements, which represents a possible shift in common indications. Notably, BMI and indication for PEG placement cannot be used to predict success or failure of the procedure. Procedural characteristics such as anesthesia type or location of placement were significantly associated with PEG success. Interestingly, endoscopist characteristics such as whether the endoscopist was a service or private attending had a significant effect on PEG success rate. This may indicate that success of PEG placement relies less on patient anatomy than on endoscopist and procedural characteristics. As this was a single-center retrospective study, there remains a need for further studies on this topic with a goal to identify better risk stratification for successful PEG placement.

Article information Continued