Jejunostomy tubeFeeding catheterEsophagectomyEsophageal cancerNational Readmissions Database
Esophagectomy remains one of the most complex surgical procedures commonly performed today, with historical complication rates as high as 80%.1 Because of the potential for serious morbidity associated with an anastomotic leak, patients traditionally have been kept nil-per-os (NPO) for extended periods of time after esophagectomy. Jejunostomy tubes (j-tubes), inserted either percutaneously or surgically at the time of esophagectomy, are commonly used to provide enteral nutrition in the early postoperative period. Compared to parenteral feeding, early enteral feeding after esophagectomy improves nutritional status, lowers costs, and decreases the rate of infectious and thrombotic complications.2
The use of feeding catheters, however, is associated with frequent catheter-specific complications, such as occlusion, dislodgement, and superficial skin infection. Indeed, tube-related complications are the leading cause of return visits to the emergency department following esophagectomy, increasing patient distress and costs.3 Less commonly, tube-related complications can be serious and require urgent reoperation. Patients who suffer such complications and are readmitted after esophagectomy are at significantly higher risk of short-term and long-term mortality.4,5 Conversely, as esophagectomy techniques improve, direct and early oral feeding without supplemental enteral access has shown to be a feasible alternative to j-tube placement for some patients undergoing esophagectomy.6,7 As such, the widespread placement of j-tubes during esophagectomy may not be as valuable as it was once thought to be.
Using a nationally representative and longitudinal database, we aimed to understand the associations of j-tube placement during esophagectomy on measures of resource utilization in terms of readmissions, morbidity, and in-hospital mortality. We hypothesized that routine feeding jejunostomy placement during or before esophagectomy may be associated with increased readmissions and healthcare utilization.
Materials and methods
Approval and permissions
The present study was approved by the Institutional Review Board of Thomas Jefferson University.
The National Readmissions Database (NRD) was developed through a Federal-State-Industry partnership sponsored by the Agency for Healthcare Research and Quality (AHRQ). This data set is constructed using data from all hospitalizations occurring in each of the 27 contributing states. Unweighted, it contains data from approximately 17 million discharges each year from 28 geographically dispersed states. The NRD is designed to be used with discharge weights (DISCWT) to create national estimates of readmissions in the United States; weighted, it estimates roughly 36 million discharges.8 The methodology used to create these discharge weights is described by the AHRQ as first stratifying each hospital in the sample according to key variables of size and location, and then creating weights for each stratum by dividing the number of discharges nationally by the number represented in the NRD within each stratum. Weighted estimates are commonly used in all studies involving the NRD to create a nationally representative sample.9, 10, 11
The NRD includes data on individuals of all ages and payers. Over 100 clinical and nonclinical variables for each hospital stay are available for study, including demographics (i.e., age, sex, quartile of median household income, and urban/rural location of the patient’s residence), payer (e.g., Medicare, Medicaid, private, uninsured), International Classification of Diseases, Clinical Modification 9th Edition diagnosis and procedure codes, timing between admissions, duration of stay of the index operation, identification of transfers, same-day stays, identification of patients who reside within the state, and hospital characteristics (size, teaching status, ownership status, urban-rural designation). Disease-specific information, such as cancer staging criterion, are unavailable. Unique patient identifiers are only valid within the same calendar year and cannot be used to track patients across years. NRD-provided weights were applied to generate national estimates.
Using the NRD, we identified patients within the database with esophageal cancer who underwent esophagectomy between 2010 and 2015. The diagnosis of esophageal cancer was defined according to International Classification of Diseases, 9th edition, Clinical Modification (ICD-9-CM) diagnosis codes corresponding to esophageal malignancy (150.0–150.9), esophageal carcinoma in situ (230.1), or malignancy of the gastric cardia (151.0). To identify those undergoing esophagectomy, ICD-9-CM procedure codes for esophagectomy (42.4–42.43) and intrathoracic esophagogastrostomy (42.5–42.52) were used. Subjects were stratified according to the presence of a jejunostomy tube during or before the index admission defined by ICD-9-CM procedure codes 46.31, 46.32, 46.39 or ICD-9-CM diagnosis codes V44.1, V44.4, or V44.9 (Fig. 1). This includes j-tubes placed at the time of esophagectomy and prior to admission for esophagectomy, i.e. during neoadjuvant treatment. Additional procedure and diagnosis codes were used to identify comorbidities, complications, and procedural details.
The NRD does not track data for patients between different years, which means events that occur in the latter part of the year may be lost. In order to ensure a full six months of follow-up after surgery for all patients in the study, we only included patients whose admissions for esophagectomy occurred in the first six months of the year.
Demographic variables in the NRD that were assessed include age, sex, payer, income quartile based upon median household incomes in the patient’s zip code ($1–$41,999; $42,000–$51,999; $52,000–$67,999; $68,000+), Charlson comorbidity index, admission status (elective or emergent), and hospital ownership, size, and teaching status (urban non-teaching, urban teaching, rural). Baseline severity of illness and risk of mortality were accounted for by using the custom variables “all-patient refined, diagnosis related groups severity of illness” (APR-DRG severity of illness) and “all-patient refined, diagnosis related groups risk of mortality” (APR-DRG risk of mortality), measures which assign classifications to illness burden or mortality risk by using software developed by 3 M Health Information systems for the NRD. Definitions of hospital size and cutoffs for income quartiles vary by census region. Neoadjuvant chemotherapy (for any unspecified malignancy) was identified with ICD-9 diagnosis codes V87.41 and V58.1 and procedure code 99.25 found prior to index admission; radiation therapy was identified with diagnosis codes V15.3 and V58.0 and procedure code 92.2. Infusion of parenteral nutrition was identified with procedure code 99.15.
The primary outcome was 30-day readmission. Secondary outcomes included in-hospital death, specific complications, total complications, and 180-day readmission rates stratified by presence of concomitant jejunostomy tube. Complications were identified by ICD-9 codes and included stroke, myocardial infarction, cardiac or renal failure, shock, acute bleeding, pulmonary embolism (PE), deep venous thrombosis (DVT), sepsis, pneumonia, bleeding requiring transfusion, intraabdominal abscess, and bowel obstruction. Patients that died during their initial hospitalization for esophagectomy were excluded from follow-up analysis for readmission. Potentially preventable causes for readmission were identified by principal diagnosis codes upon readmission; ICD-9 codes were grouped based upon their first three digits, and codes for chronic or pre-existing conditions such as “esophageal cancer” or “hypertension” were excluded. We also looked for major operative procedures of the gastrointestinal tract that occurred upon readmission with ICD-9 procedure codes 42–54.
Sensitivity analysis with propensity score-matching
In order to account for the possibility of treatment bias skewing our results, propensity score-matching was performed to validate our unadjusted multivariable regression model. A propensity score was generated for the likelihood of j-tube placement using a logit model including age, gender, APR DRG severity of illness subclass, insurance payer, income level, hospital ownership type, hospital teaching status, hospital bed size, elective admission status, and neoadjuvant therapy. EWJ cases were matched 1:1 to EWOJ controls with a caliper of 0.001. In our model, median bias in the matched cohort was 2.9% compared to 7% in the unmatched sample. All statistical analysis was performed in Stata v 13.0 (©StataCorp LLC, College Station, TX).
Descriptive statistics were used to report demographic, clinical, and hospital-related characteristics, readmission rates, causes for readmission, and in-hospital complications and mortality. Pearson’s chi-squared analysis was used to compare categorical factors. Multivariable Cox regression controlling for potential confounders (identified a priori) was used to estimate hazard ratios for factors associated with 30-day readmission and other secondary outcomes. NRD survey weights were used to generate national estimates.
After applying inclusion and exclusion criteria, 9082 patients who underwent esophagectomy for esophageal cancer were identified. After applying survey weights, this sample was estimated to 22,429 cases. Of these, 16,829 (75%) underwent esophagectomy with j-tube placement (EWJ) and 5600 (25%) did not have j-tubes placed at the time of or prior to esophagectomy (EWOJ).(Fig. 1)
Overall, 83.1% of patients were male with an average age of 67.3 years (SD 9.6). EWJ and EWOJ patients did not differ in age, gender, CCI category. There was a significant difference in the proportion of patients in the “extreme risk” APR-DRG risk of mortality subclass (16.4% EWOJ vs. 13.4% EWJ, p = 0.01). EWOJ patients were more likely to be insured by Medicare (EWOJ 48.7%; EWJ 46.0%), less likely to be privately insured (EWOJ 39.1%; EWJ 44.4%; p < 0.01), and more likely to be in the lowest income quartile (22.8% of vs. 20%, p < 0.01). In terms of urgency of the procedure, 92.2% of patients with j-tubes underwent elective esophagectomy versus 84.1% of patients without j-tubes (p < 0.01). Among the EWJ group, 5.4% were placed preoperatively during a separate admission within the same calendar year. Of those patients who received preoperative anti-neoplastic chemotherapy, 24.2% were EWJ patients and 15.1% were EWOJ (p < 0.01). Similarly, 22.7% and 13.6% of EWJ and EWOJ patients received radiation therapy (p=<0.01); 10.2% of EWOJ and 7.2% of EWJ patients received TPN (p < 0.01). An intrathoracic approach (i.e. for an Ivor-Lewis or McKeown esophagectomy) was utilized in 38% of EWOJ and 37.4% of EWJ cases (p = 0.72).(Table 1)
Table 1. Demographic, clinical, and hospital characteristics of patients undergoing esophagectomy for esophageal cancer, by J-tube placement (weighted for national estimates)†
|No J tube||% no J tube||J tube||% J tube||Total||% Total||p-value|
|Mean age (years)||63.6 + 10.2||63.6 + 9.4||63.7 + 9.6||0.55|
|Median age (years)||65||64||64|
|History of Chemotherapy||845||15.1||4065||24.2||4910||21.9||<0.01∗|
|History of Radiotherapy||760||13.6||3814||22.7||4574||20.4||<0.01∗|
|Income Quartile (0–25)||1280||22.8||3359||20.0||4638||20.7||<0.01∗|
|Income Quartile (25–50)||1569||28.0||4179||24.8||5748||25.6|
|Income Quartile (50–75)||1406||25.1||4441||26.4||5847||26.1|
|Income Quartile (75–100)||1280||22.8||4569||27.2||5849||26.1|
|Large urban (>1 mil)||3049||54.4||11422||67.9||14471||64.5||<0.01∗|
|Small urban (<1 mil)||2200||39.3||4511||26.8||6711||29.9|
|Hospital Bed Sizeꝉ|
|APR DRG Risk of Mortality|
|Minor Risk (Subclass I)||1751||31.3||5114||30.4||6865||30.6||0.62|
|Moderate Risk (Subclass II)||1658||29.6||5464||32.5||7122||31.8||0.06|
|Major Risk (Subclass III)||1275||22.8||4002||23.8||5277||23.5||0.42|
|Extreme Risk (Subclass IV)||916||16.4||2249||13.4||3165||14.1||0.01|
CCI: Charlson comorbidity index; TPN: total parenteral nutrition.
∗denotes statistical significance at α=0.05
†All values with fewer than 10 patients have been suppressed to protect patient identities in accordance with the AHRQ data use agreement
‡Involves a thoracoscopy or thoracotomy, i.e. Ivor-Lewis or McKeown esophagectomy
§Charlson comorbidity index
ꝉDefinitions of bed size (small, medium, large) vary according to hospital region, the urban-rural designationof the hospital, and teaching status (https://www.hcup-us.ahrq.gov/db/vars/hosp_bedsize/nrdnote.jsp).
Hospital-level demographics (Table 1)
Esophagectomies were largely performed at urban teaching hospitals (EWOJ 82.0%; EWJ 84.9%; p = 0.08) instead of non-teaching hospitals, or at hospitals that were privately owned (EWOJ 71.8%; EWJ 76.5%, p = 0.03) instead of federally owned. Most esophagectomies were performed in hospitals that were designated as large based on bed size (approximately 400 beds or more) (EWOJ 69.7%; EWJ 72.5%, p = 0.59).
In total, 1411 EWOJ patients (25.2%) and 4091 (24.3%) EWJ patients had an unplanned readmission within 180 days of discharge from their index admission for esophagectomy (p = 0.37); 15.2% of patients without j-tubes and 14% of patients with j-tubes were readmitted within 30 days (p = 0.16). A minority of all patients were brought in for planned, elective readmissions (EWOJ 5.1%; EWJ 4.8%, p = 0.61). Patients were readmitted to a different hospital than their index hospitalization in 5.4% of EWOJ and 7.1% of EWJ cases (p = 0.002); 5.3% and 4.4% of all EWOJ and EWJ cases underwent a major operative procedure upon readmission (p = 0.68). The median number of days elapsed between index admission and readmission was 23 days in patients without j-tubes and 22 in patients with j-tubes (p = 0.10). Fig. 2 shows the cumulative rate of unplanned readmissions over time for EWJ, EWOJ, and all patients.
The inpatient mortality rate was 5.8% for EWOJ patients and 3.8% for EWJ patients (p = 0.04); 3.5% and 2.7% died upon readmission (p = 0.61), respectively. A complication occurred in 23.6% of EWOJ and 25.5% of EWJ (p = 0.37). Postoperative small bowel obstruction, volvulus, or intussusception occurred in 0.6% of EWOJ and 0.8% of EWJ (p = 0.69). Mean length of stay was slightly lower in the j-tube cohort (14.6 ± 12.1 vs. 16.4 ± 16.1, p = 0.21). After j-tube placement, 2.3% of patients with j-tubes still received TPN during a subsequent admission. Among patients with j-tubes, 58% were discharged with homecare compared to 40.3% of EWOJ (p < 0.001); 13.8% vs. 14.6% were discharged to a skilled nursing facility (SNF) or another care facility (p = 0.81). Patients discharged to a SNF after esophagectomy more frequently died upon readmission than patients discharged home (6.3% vs. 2.4%, p < 0.001).
Multivariable Cox regression
In a Cox proportional hazards regression accounting for potential confounders such as age, gender, diagnosis of malnutrition, history of chemotherapy, zip code income quartile, and APR-DRG risk of mortality and morbidity, we found that j-tube placement was not associated with readmissions at 30 days (HR 0.90 [0.77–1.05]) and 180 days (HR 0.94 [0.79–1.10]) after discharge (Fig. 3). EWJ patients were less likely to die during readmission than their EWOJ counterparts (HR 0.72 [0.55–0.93]). EWJ was not associated with elective readmissions (HR 0.83 [0.66–1.05]) or readmissions with major operative procedures (HR 0.85 [0.67–1.09]) relative to EWOJ.(Table 2)
Table 2. Results of multivariable cox regressions for time-sensitive variables (weighted).
|Event||Hazard Ratio with J-tube||95% CI|
|180-day elective readmission||0.83||[0.66,||1.05]|
|Death during readmission||0.72||[0.52,||0.99]|
|Readmission to different hospital||1.18||[0.99,||1.40]|
|Readmission with major operation||0.85||[0.67,||1.09]|
Reasons for readmission
The most common reasons for readmission after esophagectomy are listed in Table 3. The most common primary diagnosis codes for readmission among patients with j-tubes were for unspecified digestive complications (n = 127, 7.32% of readmissions in those with j-tubes), pneumonia (n = 87, 5.02% of readmissions in those with j-tubes), and unspecified esophageal disorders (n = 83, 4.79% of readmissions in those with j-tubes). Feeding tube-related complications were the ninth most common readmission diagnosis, with 3.2% of readmissions among those with j-tubes being related to such complications. Specifically, these included mechanical obstruction (1.4%), enterostomy-site infection (0.6%), and unspecified causes (1.1%). Failure-to-thrive was not among the most common reasons for readmission.(Table 3)
Table 3. Most common potentially preventable readmission diagnoses.
|Primary Readmission Diagnosis||EWOJ||% of EWOJ Readmissions (n = 1417)||EWJ||% of EWJ Readmissions (n = 4108)||Total||P|
|Unspecified GI complications||96||6.7||300||7.3||396||0.71|
|Unspecified postoperative infection||51||3.6||192||4.7||243||0.32|
|Nausea and vomiting||57||4.0||121||2.9||178||0.49|
|Atrial fibrillation and flutter||24||1.7||123||3.0||147||0.66|
|Colostomy and enterostomy complications||21||1.5||133||3.2||154||0.01|
|Infection of enterostomy site||–||–||26||0.6|
|Other enterostomy complication||–||–||49||1.1|
EWOJ: esophagectomy without j-tube.
EWJ: esophagectomy with j-tube.
Enterostomy complications: i.e. j-tube and feeding catheter related complications.
∗All values with fewer than 10 patients have been suppressed to protect patient identities in accordance with the AHRQ data use agreement.
After propensity score matching, 4194 patients remained (2097 EWJ, 2097 EWOJ). Both groups had no statistically significant differences in gender, insurance payer, income level, hospital ownership type, hospital teaching status, hospital bed size, elective admission status, malnutrition, or neoadjuvant therapy. EWOJ patients were slightly older (mean 68.3 vs. 64.4, p = 0.05) and more likely to be in the highest APR-DRG severity of illness subclass (23.7% vs. 19.0%, p < 0.001).
In this subset of matched patients, inpatient mortality remained lower among EWJ patients than EWOJ patients (5.3% vs. 4%, p = 0.048). There was no difference between EWJ and EWOJ groups in terms of 30-day readmissions (15.2% vs 13.5%, p = 0.20), 180-day readmissions (25.0% vs. 23.3%, p = 0.29), or death during 180-day readmission (3.6% vs. 2.6%, p = 0.14). Having a j-tube did not significantly affect 180-day readmission rates (HR 0.87 [95%CI 0.74–1.02] but a higher APR-DRG illness subclass was predictive of such readmissions (HR 1.5 [95%CI 1.25–1.86]). However, EWJ patients had lower rates of pneumonia (11.6% vs. 14.4%, p = 0.008), shock (4.2% vs. 6.5%, p = 0.001), and postoperative sepsis (8.6% vs. 10.2%, p = 0.01), although the groups did not differ with regards to bleeding, DVT, PE, myocardial infarction, or respiratory failure.
Our study of readmissions and morbidity following esophagectomy using a nationally representative database shows that three out of four esophagectomies are still performed with jejunostomy tube placement. The use of j-tubes during esophagectomy did not vary significantly with basic patient-level demographics such as age, sex, or race, but was more commonly performed in patients treated at urban teaching hospitals, in those who underwent elective esophagectomy, in those who were privately insured or with higher income, and in patients who had received chemotherapy. With regards to our primary endpoint, a quarter of patients undergoing esophagectomy are non-electively readmitted within 180 days, the majority of which occur within the first month after discharge. J-tube placement was not significantly associated with readmission within 30 or 180 days. However, j-tubes may be associated with decreased inpatient mortality and infectious complications, and are linked with a higher rate of discharge to home rather than to a skilled nursing facility or other secondary location.
This evidence suggests that, contrary to our expectations, placement of j-tubes may not significantly increase inpatient readmission despite the added risk of experiencing tube-related complications. It is likely that many of the complications that arise from j-tube placement are not serious enough to warrant inpatient admission. While the incidence of tube-related complications among those with j-tubes was previously shown to be 30–51%,17,18 the vast majority of these (up to 90%) were limited to mild complications such as superficial cellulitis and tube dislodgment.19 These complications are typically managed over the phone or in the outpatient setting, with only 3–4% warranting operative intervention.20 Severe j-tube related complications are exceedingly rare but can be life-threatening, including small bowel necrosis, intestinal torsion, mechanical bowel obstruction, and intussusception.21,22 In our study, complications directly related to j-tube placement were not within the top five groups of diagnoses upon readmission, and those more severe complications were not commonly seen. Additionally, feeding through sites distal to the pylorus is associated with decreased risks of certain complications such as aspiration and dehydration.23 Although we found that inpatient resource utilization is not higher in patients with j-tubes, it is possible that patients with j-tubes may utilize outpatient resources and emergency department (ED) care more frequently. Kidane et al. found that feeding tube-related problems were the most common cause (39%) of returning to the ED within the first year after esophagectomy.3 Although these less severe complications problems warranting ED visits may be frequent and troublesome, these visits alone are not captured in the NRD. Nonetheless, j-tubes are not associated with an increase in inpatient healthcare utilization, but are associated with a higher rate of discharge with homecare, allowing patients to return home instead of requiring further stays at rehab centers or skilled nursing facilities.
The reason for the mortality benefit observed with j-tube placement in this cohort is less clear due to limitations of the data, but this effect persists even when accounting for different baseline risks (as indicated by APR-DRG mortality subclass) between both groups, and in both of our multivariable proportional hazards model and the propensity-matched cohort. This mortality benefit has been corroborated by a study performed by Lorimer et al. using the Surveillance Epidemiology and End Results (SEER) database, which found that the majority of their patients – all 65 years of age and older – received j-tubes during esophagectomy, and that mortality among those with j-tubes was lower at 30, 60, and 90 days after esophagectomy with a correspondingly lower length of stay.12 Based on their data, the authors recommended routine j-tube placement during esophagectomy as the standard of care. We can similarly describe that j-tubes are associated with a mortality benefit without an increase in inpatient resource utilization, although a causal effect cannot be assumed; while we may presume that patients with j-tubes were likely to receive some form of supplemental enteral nutrition, we can neither confirm this nor infer the type, duration, or frequency of enteral feeding actually given. Nonetheless, the benefits of perioperative enteral nutrition in patients undergoing major surgery have been extensively described in the literature,13, 14, 15, 16 and are a plausible reason for the mortality benefit observed here. Evidence from randomized controlled trials have specifically demonstrated that early postoperative enteral nutrition is safe and hastens clinical recovery after major surgery for gastrointestinal malignancies as a whole when compared to providing parenteral or delayed oral nutrition.7,24 In particular, early enteral feeding after esophagectomy has been associated with a lower rate of infectious complications (postoperative sepsis, pneumonia),2 which may explain our results and the mortality benefit observed in the propensity-matched EWJ group. Propensity score matching eliminates some (but not all) of the “healthy user” effect, wherein patients in either group could have certain characteristics or conditions (i.e. malnutrition or advanced cancer) that make them more likely to receive nutritional support.
Some may see the similar rate of readmissions between both cohorts and adopt a glass-half-full philosophy, saying that abstaining from j-tube placement is feasible and that their postoperative mortality risk is more affected by their baseline risk factors than whether they have a j-tube. Indeed, recent literature has suggested that esophagectomy can be performed without j-tube placement or any other kind of supplemental nutrition. Such enhanced recovery protocols prioritize early direct oral feeding and early mobility, and have been carried out in small populations with success.25 Prospective, non-randomized trials have both demonstrated that early, direct oral feeding after esophagectomy is not only feasible, but also associated with a shorter return of bowel function and length of stay.6,7,26 Certainly, early oral feeding appears to be feasible and may improve recovery in certain patients, especially as esophagectomy techniques continue to improve; however, patients who did not receive a j-tube may have been excluded from placement due to other factors that are not captured by the NRD. Only a minority of this EWOJ group were coded as having received TPN, indicating an alternative method of feeding, such as nasoduodenal tube feeding. Both TPN and nasoduodenal feeding have been shown to be inferior to j-tube feeding with respect to infectious complications, functional ability, or durability of access.27,28 Use of these techniques may be part of the reason why our study cohort without j-tubes seems to have inferior survival. Of note, our analysis did not show a significant difference in the rate of sepsis or pneumonia between cohorts. Ultimately, esophagectomies are still morbid operations with significant rates of 30- and 180-day complications, with j-tube complications being relatively minor events. Omission of j-tubes should still be considered judiciously and may be best reserved for those with lower baseline mortality risk.
The major limitations of this study are largely in relation to using a claims-based inpatient database to answer clinical questions. In general, we are reliant on accurate and consistent coding. This random error may lead us to over- or underestimate the rate of certain complications, and to miscode certain factors such as elective status of admission, which is lower in our cohort than has been previously described in other esophagectomy studies.29 Therefore, we can only accept the inherent selection biases that certain surgeons may have. Our propensity score matching attempts to account for these hidden biases, but can only do so to a limited extent.
Most importantly, the NRD lacks certain data regarding the actual use and duration of perioperative enteral nutrition. We may assume that most patients who receive a j-tube may receive some period of enteral supplementation, but it is impossible to know the duration of this treatment. Similarly, the code for tube-related complications may also capture complications related to intestinal stomas, including ileostomies and colostomies.30 As a result, it is probably more apt to focus on the incidence (or lack thereof) of readmissions rather than assuming that differences in complications are due to disparities in nutritional status between groups. Our results may also underestimate the full extent of healthcare utilization in both groups, as the NRD is unable to capture emergency department visits and hospital stays under “observational” status. This also means that we may be underestimating the rate of adjuvant and neoadjuvant therapy, much of which may happen in the outpatient setting at non-index facilities.
The NRD also lacks cancer-specific variables such as staging or pathologic subtype, which are factors that may influence the decision to place a j-tube and could affect long-term survival and readmission rates. Furthermore, reasons for inpatient mortality cannot be ascertained from this database. We can only correlate between certain complications and death, but cannot infer causality between the two. Similarly, although we can identify patients with a history of chemo- or radiotherapy, we cannot be certain that these treatments were for esophageal cancer. Lastly, CPT codes for minimally invasive esophagectomies were not put into widespread use until 2018, making identification of minimally invasive cases nearly impossible with this claims database.
Perioperative nutritional support during esophagectomy remains the clinical standard of care. Placement of jejunal feeding catheters is associated with improved survival after esophagectomy and increased rates of discharge to home without a corresponding increase in readmissions (Fig. 4). Careful consideration should still be taken when deciding not to place a j-tube before or during esophagectomy, although the continual adoption and improvement of minimally invasive techniques may change this in the future. Further study should be in the form of a randomized, controlled trial selecting for j-tube placement during esophagectomy.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of competing interest
The authors of this work do not have any conflicts of interest to disclose.