Elimination of Routine Feeding Jejunostomy After Esophagectomy


Feeding jejunostomy is frequently used to ensure nutritional intake after esophagectomy. Early return to diet is demonstrated to enhance recovery in major abdominal surgery. Early oral feeding is safe and effective in recent randomized controlled trials in esophagectomy. This study assesses the implications of eliminating the insertion of jejunostomy after esophagectomy.


A retrospective study was undertaken between 2014 and 2017 with follow-up over the first year. Fifty patients did not have a jejunostomy, compared with 46 patients who had conventional treatment. Outcomes measured included change in relative weight and body mass over 1 year, complications, and nutritional reinterventions.


Median weight loss at 1 year was 10.7 kg (range, –8 to 55.6) whereas median percent weight loss was 12% (range, –10.1% to 39.2%). Patients without jejunostomy lost more weight during the first month (P = .002). Thereafter, at 6 of 12 months, there were no differences in actual or relative weight loss. Obese patients lost more weight in the group without jejunostomy compared with those who had it (9.9 versus 5 kg; P = .004). This effect was not seen in normal or overweight patients. Complications were similar, whereas leaks were more common in the jejunostomy group (15.2% versus 2%; P = .019). Nutritional reinterventions were similar during index admission and subsequent readmissions (7 versus 5 patients; P = .640).


Routine jejunostomy use delays rather than prevents weight loss after esophagectomy. Oral route nutrition allows patients to maintain sufficient nutrition and does not increase complications or requirement for nutritional interventions after surgery. Routine use of jejunostomy may not be required in modern practice.

Esophagectomy is a morbid procedure associated with multiple complications that prolong hospital stay and considerably affect recovery.1 The use of feeding jejunostomies is considered the reference standard and is typically a prerequisite in postesophagectomy care. The application of delayed return to oral feeding, in deference to the perceived increase in rate of anastomotic dehiscence and pulmonary complications, is widely applied.2 However, this tenet of esophagectomy care is now being challenged.3

With the advent of minimally invasive esophageal surgery and enhanced recovery principles, there has been a demonstrable shortening of hospital stay, better quality of life, and faster return to function.4,5 An important principle of enhanced recovery protocols is early commencement of diet. In other gastrointestinal surgeries, enhanced recovery practices include early return to diet promoting earlier return of gut function and overall recovery.6,7 Within upper-gastrointestinal surgery, early oral intake has been demonstrated to reduce complications and hospital stay.8 It remains contentious in esophagectomy care as to whether early oral feeding should occur. Nutritional interventions were previously common after surgery, predominantly for complications or when oral intake was insufficient.9 Recently, early feeding was demonstrated to be applicable and safe in esophagectomy with sufficient calorie intake.10,11 This implies that sole, direct feeding after esophagectomy is feasible.

Weight loss occurs after esophagectomy irrespective of the choice of feeding route.12 In the first month, up to 82% of patients lose weight.13 Furthermore, only a fraction of patients will regain baseline weight; up to 10% body weight loss occurs in the initial 6 months.14

This study examined the experience of eliminating routine insertion of jejunostomies after esophagectomy with direct oral feeding compared with traditional jejunostomy practice. Outcomes studied included postoperative morbidity, nutritional interventions, and the effect on weight in the first year after esophagectomy.

Patients and Methods

This was a single-center retrospective analysis of consecutive patients undergoing esophagectomy for cancer over 36 months between January 2014 and December 2017. During this time, routine insertion of a feeding jejunostomy was discontinued, as was pyloric drainage. This was a practice change among surgeons performing esophagectomy. This rationale emanated from observed increases in unexpected returns to clinic and emergency departments, as well as complications from jejunostomies requiring intervention. Patients not completing 1-year follow-up were excluded (death from any cause or loss to follow-up). All patients had 90-day follow-up from complications as per Esophageal Complication Collaborative Group standards.15 All patient data were obtained from a prospectively maintained database. This study was approved by the Research Ethics Board of the University Health Network.

Postoperative Nutritional Protocol

This study incorporated 2 patient groups following separate feeding protocols. Both groups adhered to an overarching strategy regarding analgesia (routine avoidance of epidurals), early mobilization (postoperative day [POD] 1), drain removal (nasogastric [NG] or chest tube), and nursing care.

Patients in the feeding jejunostomy (FJ) group received enteral nutrition commencing on POD 1 with graduated increase until calorie requirements were met and tolerated. Patients commenced oral diet between POD 4 and 5, gradually increasing from liquid consistency to a postesophagectomy diet. These patients continued supplemental enteral feeding at home until the first review in clinic at week 2 to 3 after discharge. At this point, the jejunostomy was removed if oral nutrition was deemed sufficient. Prolonged jejunostomy feeding was defined as the requirement for enteral feeding at greater than 30 days after resection. Most patients in this group had pyloric drainage, as was standard practice at the time.

For patients without a feeding jejunostomy (NFJ), oral feeding was feeding was commenced upon NG drain removal (the determination was based on the absence of conduit distension upon chest x-ray and low-volume NG output [<200 mL/24 h]). Once water was tolerated, progression from clear fluids to a postesophagectomy diet was initiated over subsequent days. Evidence of delayed gastric emptying (for radiological, conduit distension on chest x-ray; for clinical, regurgitation or emesis) prompted cessation and consideration for pyloric intervention. Parenteral nutrition was used from POD 1 until a postesophagectomy diet was initiated or if a chyle leak was identified in either group. During this period, there was an institutional change in practice in which pyloric drainage was omitted.

Surgical Technique

A thoracolaparoscopic esophagectomy was performed when feasible. For cancer of the lower esophagus and gastroesophageal junction, an abdominal D1+ lymphadenectomy and gastric conduit formation was performed laparoscopically followed by thorascopic lymphadenectomy and end-to-side anastomosis in a left lateral or semiprone position. This anastomosis was created preferentially using a 25-mm EEA circular stapler (Medtronic, Dublin, Ireland). For cancers of the mid or proximal esophagus, thorascopic lymphadenectomy (as described earlier) and esophageal mobilization, followed by laparoscopic abdominal lymphadenectomy, gastric conduit formation, and finally, a left cervical incision and modified Collard semimechanical anastomosis using a linear GIA surgical stapler (Medtronic). A jejunostomy was placed during the abdominal phase of surgery in the FJ group. Pyloric drainage was performed in the earlier portion of the series, but it was subsequently omitted owing to the emerging evidence base. The integrity of the anastomosis in each case was assessed during surgery with endoscopy. An NG tube was subsequently placed.


Primary outcomes were (1) nutritional reinterventions (prolonged jejunal tube feeding, recommencement of tube feeding or total parenteral nutrition (TPN), and readmissions for feeding interventions), and (2) body mass changes over the first year after surgery. Body weight was measured in kilograms and body mass index (BMI) was calculated in kilograms per square meter and classified as normal weight (BMI ≤24.9), overweight (BMI ≥25 and ≤29.9), or obese (≥30). Absolute and relative weight change in kilograms and percentage were calculated at 1, 6, and 12 months after surgery. Secondary outcomes included complications graded according to the Clavien-Dindo classification16 and the comprehensive complication index.17

Statistical Analysis

Summary data are presented as median and interquartile range or frequency and number of patients (percentage). Statistical analysis was performed using chi-square or Fisher exact tests for categorical variables. Comparisons between groups were done using Student t test or Mann-Whitney U test. Data analysis used parametric or nonparametric statistics, depending on data distribution adherence to normality. P less than .05 was used as significance.


Of 126 consecutive patients undergoing esophagectomy over the study period, 96 were available for final analysis. Of the excluded patients, 10 underwent jejunal, colonic interposition or other complex reconstruction, whereas the remaining 20 did not have complete measurements noted over the follow-up period. A proportion of these patients travelled from northern Ontario for surgery. Follow-up is organized by local oncology services. Median age for the study group was 64.0 years (range, 35-86 years); 73 were male (76%). Neoadjuvant treatment was received by 67 (69.8%). The vast majority of tumors (n = 78; 81.1%) were located in the distal esophagus and gastroesophageal junction. Baseline characteristics were essentially similar between groups (Table 1). Pyloric drainage was performed more frequently in the feeding jejunostomy group (89.1% versus 14%; P < .001).

Table 1. Patient Clinicopathological Demographicsa

VariablesCohort (n = 96)Feeding Jejunostomy (n = 46)No Feeding Jejunostomy (n = 50)P
Age (range)64.0 (35-86)66.0 (40-86)63.0 (35-78).171
Sex (%).992
 Male73(76)35 (76.1)38 (76)
 Female23(24)11 (23.9)12 (24)
American Society of Anesthesiologists classificationIII (II-IV)III (III-IV)III (II-IV).253
 Minimally invasive7838 (82.6)40 (80)
 Hybrid303 (6)
 Open158 (17.4)7 (14)
Pyloric drainage48 (50)41 (89.1)7 (14)<.001
 Adenocarcinoma71 (74.0)29 (63.0)42 (84)
 Squamous cell20 (20.8)14 (30.4)6 (12)
 Other5 (5.2)3 (6.5)2 (4)
Tumor location.085
 Proximal1 (1.0)01 (2)
 Middle17 (17.1)10 (21.7)7 (14)
 Distal33 (34.4)20 (43.5)13 (26)
 Junction45 (46.9)16 (34.8)29 (58)
Location of anastomosis.916
 Thoracic90 (93.8)43 (93.5)47 (94)
 Cervical6 (6.2)3 (6.5)3 (6)
Neoadjuvant therapy.755
 Surgery alone29 (30.2)14 (30.4)15 (30)
 Chemoradiation54 (56.3)27 (58.7)27 (54)
 Chemotherapy13 (13.5)5 (10.9)8 (16)


Data are shown as n (%).

Weight and Body Mass Change Over 1 Year

Body mass index at time of admission for surgery in the entire group was 26.5 kg/m2 (range, 16.5-52.3 kg/m2). There was no difference in BMI between groups at the time of surgery. At 1, 6, and 12 months after esophagectomy, median BMI for the study group was 25.2 kg/m2 (range, 14.9-43.5 kg/m2), 23.8 kg/m2 (range, 14.7-39.7 kg/m2), and 23.3 kg/m2 (range 14.3-40.3 kg/m2), respectively.

Cumulative weight loss and BMI alteration are demonstrated in Figure 1. Actual median weight loss at 1 year after surgery was 10.7 kg (range, –8 to 55.6 kg), whereas the total percent weight loss in this study at this time point was 12% (range, –10.1% to 39.2%). At 1 month after surgery, NFJ patients lost a median of 5.8 kg (range, –3.0 to 18.8 kg), compared with 3.6 kg (range, –3.2 to 14.0 kg) in patients receiving tube feeding (P = .002). Thereafter, both groups continued to lose a variable amount of weight at a much-reduced rate with no significant differences (Table 2). Patients using the oral route exclusively for nutrition in the first month lost more weight, but the 2 groups demonstrated a similar pattern of weight change afterward.

Table 2. Body Mass and Weight Changes Over the First Year After Esophagectomy

Cohort (n = 96)Feeding Jejunostomy (n = 46)No Feeding Jejunostomy (n = 50)P
 BMI category, kg/m2.214
 ≤24.933 (34.4)19 (41.3)14 (28)
 25-29.937 (38.5)18 (39.1)19 (38)
 ≥3026 (27.1)9 (19.4)17 (34)
 Diagnosis, mo26.8 (16.5 to 52.3)26.0 (16.5 to 44)27.5 (19.6 to 52.3).069
 125.2 (14.9 to 43.5)24.4 (14.9 to 41.7)25.9 (16.9 to 43.5).408
 623.8 (14.7 to 39.7)23.2 (14.7 to 39.2)24.9 (16.4 to 39.7).107
 1223.3 (14.3 to 40.3)22.3 (14.3 to 40.2)25.5 (15.9 to 40.3).021
Weight loss, kg
 Surgery to 1 mo4.8 (–3.2 to 18.8)3.6 (–3.2 to 14.0)5.8 (–3.0 to 18.8).002
 1-6 mo3.75 (–7.1 to 26.1)4.05 (–7.1 to 20.3)1.45 (–7.0 to 26.1).814
 6-12 mo0.9 (–17.3 to 20.5)1.7 (–17.3 to 20.5)0.6 (–3.4 to 20.1).509

BMI, body mass index.

Obesity (>30 kg/m2) was present at the time of surgery in 26 of the FJ group (27.1%) and 17 of the NFJ group (24%). Examining actual weight loss at 1 month after surgery by weight class, it is evident that patients classed as obese without a feeding tube lost more weight compared with those who received enteral feeding (9.9 kg [range, 3.4-18.8 kg] versus 5 kg [range, 1.4-11.3 kg]; P = .004). This difference was not seen for the overweight class (NFJ: 4.35 kg [range, –3 to 14.8 kg] versus FJ: –3.35 [range, –3.2 to 7.5 kg]; P = .214) and normal weight class (NFJ: 4.4 kg [range, –0.7 to 14.5 kg] versus FJ: 3.7 kg [range, –0.5 to 14 kg]; P = .377). The changes in weight at 6 and 12 months by body mass did not differ (Figure 2).


The frequency of complications was similar between groups: 19 patients in the FJ group (41.3%) and 23 in the NFJ (46%) developing a complication (P = .640) (Table 3). Similarly, there was no difference in the severity of complication based on the Clavien-Dindo classification (P = .723). Rates of pneumonia and readmission to intensive care for respiratory support were similarly low. Of note, anastomotic leakage was found to be lower in the no-feeding jejunostomy group (1 of 50 [2%] versus 7 of 46 [15.2%]; P = .002).

Table 3. Surgical Complications

VariablesFeeding Jejunostomy (n = 46)No Feeding Jejunostomy (n = 50)P
Total complications19 (41.3)23 (46).640
Clavien-Dindo classification.723
 I2 (4.3)2 (4.2)
 II5 (10.9)9 (18.8)
 III10 (21.7)8 (16.7)
 IV2 (4.3)4 (8.3)
Comprehensive complication index score27.9 (8.7-79)26.2 (8.7-54.2).652
Length of stay11 (6-56)8.5 (5-69).055
Anastomotic leak7 (15.2)1 (2).019
Chyle leak2 (4.3)4 (8).460
Pneumonia3 (6.5)2 (4).579
Respiratory failure3 (6.5)2 (4).579

With respect to jejunostomy complications, 1 patient in the FJ group required readmission and urgent surgical intervention for bowel obstruction at 94 days. The jejunostomy had been removed 50 days earlier. Most issues arose from tubes dislodging, requiring reinsertion, or wound or skin irritation.

Nutritional Reinterventions

During primary admission, a single patient in the NFJ group required insertion of a jejunostomy for respiratory failure owing to pneumonia. Total parenteral nutrition was started in 5 patients in the FJ group for management of chyle leakage in 2 patients and in 3 for poorly tolerated enteral feeding. Prolonged TPN (>5 days) was required in 3 patients in the NFJ group for symptoms of delayed gastric emptying, whereas TPN was initiated for management of chyle leaks in 4 patients in this group.

A total of 12 of 96 readmissions occurred in this study for nutritional reasons (12.5%), as demonstrated in Table 4, with no differences between groups. There were 13 endoscopic interventions performed in the FJ group (28.3%) and a further 13 in the NFJ group (26.0%) in the ambulatory setting to ameliorate dysphagia resulting from anastomotic stenosis or fullness or regurgitation caused by delayed gastric emptying. Unplanned visits occurred in 5 patients who presented to clinic for management of jejunostomy tube change or wound issues in the FJ group, and the single patient who required insertion of a feeding tube in the NFJ group requested a tube change owing to leak or skin irritation. Patients reporting symptoms suggestive of delayed gastric emptying or radiological evidence of a distended conduit underwent dilatation and or botulinum toxin injection of the pylorus. In the FJ group, this was 3 of 46 patients (6.5%), compared with 8 of 50 patients (16.5%) in the NFJ group (P = .145).

Table 4. Postoperative Nutritional Interventions

Feeding Jejunostomy (n = 46)No Feeding Jejunostomy (n = 50)P
Primary admission
 Oral diet commencement, d (range)4 (3-20)3 (2-21).008
 Jejunostomy insertion46 (100)1 (2)a
 TPN commencementb5 (10.9)7 (14).767
 Chyle leak2 (4.3)4 (8).460
 Delayed gastric emptying03 (6)<.001
After discharge
 Readmission7 (15.2)5 (10).64
 Delayed chyle leak01
 Aspiration pneumonia21
 Conduit revision/hernia11
 Bowel obstructionc10
 Failure to thrive31
Ambulatory interventions
 Anastomotic dilatation10 (21.7)5 (10).114
 Gastric emptying
Pyloric intervention
3 (6.5)8 (16).145
 Jejunostomy issued5 (10.7)1 (2).128


Feeding jejunostomy placed for prolonged intubation due to respiratory failure;b

Total parenteral nutrition (TPN) commencement included prolonged TPN use in the no-feeding jejunostomy group;c

Obstruction at jejunostomy site;d

Jejunostomy tube change or leak.


This study demonstrates an early experience in eliminating routine insertion of feeding jejunostomy at esophagectomy. We demonstrate evidence of increased initial weight loss in patients without jejunostomy and that this occurred disproportionately in obese patients. After 1 month, there was no difference between those who received enteral feeding and those who did not with respect to weight loss or reduction in BMI. Similar to findings with complications and readmissions, no differences were observed.

Esophagectomy has a considerable effect on nutritional intake ability. It is akin to the restrictive and metabolic effects of bariatric procedures in which weight loss is the desired effect. Weight loss after esophagectomy is complex and occurs by many means, including reflux, absence of hunger, dysphagia, postprandial dumping, pancreatic insufficiency, and alterations in gut hormone function.18 Routine use of jejunostomies in the first few weeks delays but does not attenuate the inevitable weight loss that occurs once enteral feeding ceases.12,19 It is evident that patients who depend on the oral route lose more weight in the first month that is not readily regained. Thereafter weight loss stabilizes between groups, such that once the jejunostomy is removed, these patients lose more weight compared with the NFJ group when the 6-month mark has passed. Furthermore, we showed that most weight was lost in the first 6 months (8.8 kg; 10.9%) compared with the second 6 months (0.9 kg; 1.3%). These findings are similar to those of Berkelmans and colleagues.19 In their series of early oral feeding compared with jejunal feeding, weight loss was greatest in the initial 6 months at 6.3 kg or 8.3% body weight compared with the following 6 months, in which weight loss was 0.7 kg. Interestingly, in our study, when we examined BMI, patients considered obese (BMI ≥30 kg/m2) lost significantly more weight in the first month compared with those considered overweight or of normal weight, and that group was responsible for most of the weight loss observed in the overall cohort. Advocating for a prolonged or extended period of enteral feeding after surgery may not appear to be substantiated when examining our series.12,20

Anastomotic leakage and major morbidity are the predominant reasons for prolonged fasting after esophagectomy. In our study, rates of anastomotic leaks were lower in the NFJ group. Because this was a consecutive series of predominantly minimally invasive esophagectomies, one could argue that the latter series was further along the learning curve. It may also be explained by the use of a transoral circular stapler in a number of cases in the earlier feeding jejunostomy group that was associated with a higher leak rate. The single leak that was observed in the NFJ group occurred day 2 after surgery, and the patient was returned to surgery when a small rent at the level of the anastomosis was visible. This was a suture repaired with minimal delay in return to diet. Overall complications and their severity were equal between groups. Despite this, there was the suggestion that length of stay was shorter in the NFJ group. This may also be related to anastomotic leaks delaying discharge, or to our developing experience in enhanced recovery principles allowing for earlier discharge. Specific studies in early or immediate return to oral diet after esophagectomy have emerged and are demonstrated to be feasible, safe, and effective. In an exploratory trial examining immediate return to diet, Weijs and coworkers10 demonstrated a similar complication pattern, specifically rates of respiratory complications and a shorter intensive care unit admission and hospital stay. The requirement for NG reinsertion was similar between groups. Finally, and most important, calorie intake requirements were sufficient in the early feeding group. Most recently, the first randomized controlled trial examining immediate oral feeding in minimally invasive McKeown esophagectomies showed that early return to diet was associated with earlier return to gut function, shorter length of stay, and better short-term quality of life scores, all without compromising safety.11 The NUTRIENT II trial21 examined patients who commenced direct oral feeding on day 1 versus those fed with jejunostomy and a 5-day fast. Leak rates and other complications were similar. Functional recovery, defined as the absence of need for intravenous fluids, sufficient calorie intake, and independent mobility were similar, indicating that direct oral feeding was sufficient in the recovery of these patients. There was no difference in hospital readmissions.

Our preference is to have an NG tube placed at end of surgery, in an effort to avoid early conduit distension and guide the decision to start an oral diet irrespective of whether a feeding tube is present. Sun and colleagues22 demonstrated that the emptying time of the gastric conduit is actually improved compared with the normal stomach when liquids are consumed, which implies that delayed gastric emptying is not an overwhelming issue when commencing an early diet. In the trial mentioned earlier, the NG was placed for the initial 12 hours after surgery to detect bleeding. Reinsertion was required in 5.7% of the early-feeding group compared with 7.9% in the delayed group. In our series, reintervention for nutritional issues revolved predominantly around commencement of TPN for chyle leaks and delayed gastric emptying that was more common in the NFJ group during the index admission. This group had fewer intraoperative pyloric interventions. Most of the FJ group had intraoperative pyloric drainage, which reflected our prior practice in this series. It is difficult to state whether this affected outcomes, apart from the NFJ requiring dilatation or botulinum toxin to the pylorus. That said, the FJ group had more leaks (7 versus 1), which has been an assumed benefit of draining the conduit. More recently, it was shown that disrupting the pylorus, whether by pyloroplasty, pyloromyotomy, dilatation, or botulinum injection during surgery, carries no benefit in terms of anastomotic leak, pneumonias, or other complications.23 Complications regarding jejunostomies in the postdischarge period pertained to blockages requiring change, leakage at the site, or skin irritation. Although one might consider these minor issues, complications caused by jejunostomies are frequent, with surgical reintervention required in 2%.13,24 They are also the main reason for presentation to emergency departments.25

A limitation of this study was the inability to calculate total calorie intake at discharge in the NFJ group. All patients and their families met with a dietitian before discharge for an education session in which the focus was on high-calorie foods. We acknowledge that no data were available for long-term nutritional complications. The first postdischarge review was at 2 weeks, when troubleshooting or further advancement of diet was permitted. According to Weijs and Berkelmans,10 a calorie intake of 60% (1205 kcal) of the estimated daily requirement was achieved by most patients at POD 5. Sun and coworkers11 achieved 70% (1260 kcal) total calorie intake by the fifth day. This indicates that reasonable nutritional targets are attainable in the early postesophagectomy period. Although it is retrospective in design, this study captured data on consecutive patients during a transition in intervention that has become our standard of care. A similar study omitting jejunostomy insertion, from Kroese and colleagues,26 highlighted an experience similar to our own. Their practice was to wait 5 days before an oral diet in all patients with no apparent difference in perioperative outcomes or in the weight loss pattern at 6 months.

Esophagectomy notably affects weight and body mass in the first 12 months after surgery irrespective of the initial feeding approach. This study demonstrated that the early introduction of sole oral nutrition intake is safe. Weight loss was accentuated in the first month in patients without enteral feeding. Once jejunal feeding was discontinued, similar patterns of weight loss emerged. This approach had a minimal impact on nutritional reinterventions and complications. Such results demonstrate that eliminating routine feeding jejunostomy in esophagectomy is feasible. Close support is necessary to monitor for issues that may cause impediments to an oral diet.




Is the placement of jejunostomy tubes in patients with esophageal cancer undergoing esophagectomy associated with increased inpatient healthcare utilization? An analysis of the National Readmissions Database