Of the 324 patients who underwent PEG at the Asan Medical Center (Seoul, Korea) between January 2013 and July 2017, 27 underwent PEG in the SICU and 263 underwent the procedure in the GW. The remaining 34 patients, consisting of 32 who underwent PEG in the medical ICU and two who underwent PEG in the emergency room, were excluded from this study (Fig. 1). Both patients in the emergency room underwent PEG due to cancer progression and were not indicated for surgery. The medical records of all included patients were retrospectively reviewed, and clinical outcomes were compared between SICU and GW patients. Demographic and clinical characteristics were evaluated, including underlying diseases, body mass index (BMI), indications for PEG, complications after PEG, comorbidities, and baseline laboratory values. In addition, baseline Sequential Organ Failure Assessment (SOFA) scores and Acute Physiology and Chronic Health Evaluation (APACHE) II scores for SICU patients were evaluated. Weight and serum albumin concentrations were compared before and after PEG. If post-PEG weight and albumin level were assessed multiple times between 2 weeks and 1 month after PEG, the most recent measurements were used. PEG was performed on patients who had normal gastrointestinal function and were expected to require enteral feeding for longer than 2–3 weeks. The Institutional Review Board (IRB) of Asan Medical Center (Seoul, Korea) approved the study (IRB number, 2018-0928). The requirement for patient informed consent was waived due to the retrospective design of the study.

Of 290 PEG procedures, 289 were performed using the introducer technique and one was performed using the pull-type technique. The PEG procedure was performed by experienced gastroenterologists. The introducer (Russell) technique has been the main approach since November 2011, while pull-type PEG is usually performed for the purpose of drainage. The introducer technique was performed using a transilluminated endoscope, and the gastropexy device and trocar were used to penetrate the abdominal wall. A 15-Fr Cliny^® PEG tube (Create Medic Co., Ltd., Yokohama, Japan) was subsequently inserted directly through the abdominal wall and fixed by filling the balloon of the tube with 5 cc of saline (Fig. 2).

Of 290 patients, 279 (96.2%) were administered prophylactic antibiotic 30 min before PEG, with most receiving 2 g of cefazolin. If a patient was already receiving an antibiotic, an altered regimen was not required unless the current regimen was inappropriate for gram-positive cocci. For sedation, intravenous midazolam (0.05 mg/kg) and pethidine (25 or 50 mg) were administered before the procedure.

Acute complications were defined as complications that developed within the first 7 days of PEG insertion, and chronic complications were those that occurred after 7 days. Acute complications included bleeding, ileus, pneumoperitoneum, aspiration pneumonia, wound infection, and Mallory–Weiss tear. Chronic complications included tube obstruction, tube leakage, spontaneous tube removal, wound infection, recurrent aspiration pneumonia, and buried bumper syndrome.

Bleeding was defined as a hemorrhagic event requiring endoscopic intervention or embolization for hemostasis and reduction of >2 g/dL of hemoglobin concentration. Aspiration pneumonia was defined as newly developed symptoms, such as cough, purulent sputum, or fever after PEG, in addition to changes in chest X-ray. Diagnosis of ileus was confirmed using X-ray or computed tomography imaging indicative of ileus in addition to decreased bowel sounds with or without symptoms.

Patients were kept nil per os on the day of the PEG procedure with natural tube drainage. The next day, the PEG tube of patients with a soft abdomen and those with no abdominal discomfort was flushed with 100 mL of clear water. After the water flush, patients without abdominal pain, vomiting, or tube leakage started tube feeding with 100–200 mL of fluid.

Patients in the GW were started on tube feeding with 100 cc of formulated nutrition, with the volume incrementally increased by 50–100 cc for every meal, as tolerated until the target volume was reached. Patients in the SICU were started on continuous feeding with 20 cc/hr, with the amount increased by 10–20 cc/hr every 12–24 hr until the target volume was achieved.

Descriptive data are expressed as mean±standard deviation for normally distributed continuous variables or as median and interquartile range (IQR). Continuous variables were compared between groups using the t test or Mann–Whitey U test, and categorical variables were compared using the chi-square or Fisher’s exact test, as applicable. Multivariate logistic regression models were constructed to evaluate risk factors for complications and 30-day mortality. SPSS version 21.0 (IBM Co., Armonk, NY, USA) was used for all statistical analyses; differences with a p-value <0.05 were considered to be statistically significant.

The baseline demographic and clinical characteristics of the patients in the GW and SICU groups, including age, sex, BMI, indications for PEG insertion, comorbidities, and baseline laboratory values, are presented in Table 1. The median age of the 27 SICU patients was 66 years (IQR, 58–77 years), and their median BMI was 21.1 kg/m² (IQR, 17.8–24.0 kg/m²).

The median baseline SOFA score of the SICU patients was 4 (IQR, 3–9), and their median APACHE II score was 16 (IQR, 12–22) (Table 2). Indications for surgery in the SICU group included nine (33.3%) patients with traumatic fracture with brain hemorrhage and four (14.8%) with malignancy including mouth floor cancer, distal common bile duct cancer, hepatocellular carcinoma, and sigmoid colon cancer. Four (14.8%) patients had panperitonitis. The mean interval from surgery for panperitonitis to the PEG procedure was 36.2 days. Patients exhibited improving clinical course of panperitonitis but presented other postoperative complications, such as ventilator- associated pneumonia and cerebral infarction, which lengthened hospital stay. Seven (25.9%) patients underwent surgery for other conditions including abdominal aorta aneurysm, total hip replacement arthroplasty, infective endocarditis, kidney transplantation rejection, simultaneous pancreas kidney transplantation, and ulcerative colitis. Three patients in the SICU did not undergo any surgical intervention. One patient was admitted to the SICU due to a traffic accident and was diagnosed with brain death. A second patient was admitted to the SICU after falling and sustaining an intracerebral hemorrhage; after undergoing a tracheostomy and being treated for pneumonia, he was transferred to a local hospital. The third patient had undergone a kidney transplant and was admitted to the SICU for suspicion of acute rejection. She was treated medically for the rejection without surgical intervention.

The median interval between surgery and PEG in the SICU group was 30 days (IQR, 19–51 days), and the median follow- up period was 73 days (IQR, 21–189 days). PEG insertion was successful in 26 of the 27 patients in the SICU. PEG insertion was unsuccessful in one (3.7%) patient due to the inability of the needle to puncture the gastric wall. Repeated surgical procedures in this patient may have led to the development of postoperative anatomic distortion. This patient had undergone a subtotal colectomy for an ischemic colitis induced perforation with panperitonitis and had underlying end-stage renal failure. After the colectomy, she underwent additional anastomosis revision and ileostomy due to anastomosis site leakage. Three weeks after surgery, respiratory arrest occurred due to aspiration. PEG in this patient failed.

Post-PEG complications are summarized in Table 3. Acute complications in SICU patients included bleeding (7.4%) and ileus (11.1%), and chronic complications included aspiration pneumonia (7.4%) and tube obstruction (3.7%). The incidence rates of acute and chronic complications did not differ significantly between the GW and SICU groups.

Of the four SICU patients who developed bleeding, none required intervention to stop the bleeding. By comparison, five patients in the GW group underwent endoscopic intervention, such as clipping, coagrasper, or embolization, to stop the bleeding. Ileus was detected in three patients in the SICU group, and all patients exhibited improvement following conservative management.

Of the 27 patients in the SICU group, 12 were discharged to the GW, seven were transferred to local hospitals, and eight died during SICU admission. All 12 patients discharged to the GW were fed through PEG tubes; their median length of stay in the SICU was 53 days (IQR, 25–75 days). The median length of stay in the SICU for all 27 patients was 57 days (IQR, 28–83 days). The median interval from surgery to PEG insertion in patients with intra-abdominal surgery was 46 days (IQR, 20–71 days).

Nine patients (33.3%) were able to have their PEG tube removed and resume oral feeding. The median interval of PEG use in these nine patients was 37 days (IQR, 25–75 days).

Clinical prognosis after PEG insertion, including early mortality rate and nutritional status, is summarized in Table 4. The 30-day mortality rate in the SICU group was 14.8%, which was considerably higher than the 5.3% in the GW group. However, none of the patients in the SICU group died of procedure-related causes. Four patients in the SICU group died within 30 days of the PEG procedure. One of these patients was admitted to the SICU due to a traffic accident and was diagnosed with brain death. The second patient underwent intra-abdominal surgery after being transferred to the SICU for panperitonitis and died 15 days after PEG from aggravated ischemic colitis. A third patient with rectal cancer and multiple lung metastases was admitted to the SICU for drug-induced lung injury; this patient died of aggravated pneumonia 3 days after PEG. The fourth patient, who had ischemic colitis and underwent a right hemicolectomy and cholecystectomy, developed pneumonia and ventricular tachycardia and died of cardiac arrest 8 days after PEG insertion.

Fourteen patients in the GW group died within 30 days after PEG, including one from PEG procedure-related causes. This patient died 12 days after PEG insertion of newly developed aspiration pneumonia. Four patients died of aspiration pneumonia not related to the PEG. Other causes of death included renal failure, acute myocardial infarction, traumatic subdural hemorrhage, and sepsis due to urinary tract infection. Five patients died of unknown causes, because they were lost to follow-up.

Mean changes in weight, BMI, and serum albumin concentration from before to after PEG insertion did not differ significantly in the GW and SICU groups.