|Home | About | Journals | Submit | Contact Us | Français|
As part of our ongoing quality improvement effort, we evaluated our conventional approach to post-oesophagectomy management by comparing it to an alternative postoperative management pathway.
Medical records from 386 consecutive patients undergoing oesophagectomy with gastric conduit for oesophageal cancer or Barrett’s oesophagus with high-grade dysplasia were analysed retrospectively (July 2004 to August 2008). The conventional pathway involved a routine radiographic contrast swallow study at 5—7 days after oesophagectomy with initiation of oral intake if no leak was detected. In the alternative pathway, a feeding jejunostomy was placed for enteral feeding and used exclusively until oral intake was gradually initiated at home at 4 weeks after oesophagectomy. No contrast swallow was obtained in the alternative pathway group unless indicated by clinical suspicion of an anastomotic leak. Each group was analysed on an intention-to-treat basis with respect to anastomotic leak rates, length of hospitalisation, re-admission and other complications.
A total of 276 (72%) patients underwent conventional postoperative management, 110 (28%) followed the alternative pathway. Patient characteristics were similar in both the groups. The anastomotic leak rate was lower in the alternative pathway with three clinically significant leaks (2.7%) versus 33 in the conventional pathway (12.0%; p = 0.01). Among patients undergoing a radiographic contrast swallow examination, a false-negative rate of 5.8% was observed. The swallow study of 14 patients (5.9%) was complicated by aspiration of oral contrast. Postoperatively, 7.3% of patients suffered from pneumonia. There were no significant differences overall in postoperative pulmonary or cardiac complications associated with either pathway. Median length of hospitalisation was 2 days shorter for the alternative pathway (7 days) than the conventional pathway (9 days; p < 0.001). There was no significant difference in unplanned re-admission rates.
An alternative postoperative pathway following oesophagectomy involving delayed oral intake and avoidance of a routine contrast swallow study is associated with a shortened length of hospitalisation without a higher risk of complication after hospital discharge.
The incidence of oesophageal cancer has dramatically increased over the past several decades with an incidence of 4.6 per 100 000 per year and a death rate of 4.4 per 100 000 in the United States for the years 2001—2005. Unfortunately, the disease is still associated with an overall low 5-year survival rate of 15.8%, as stated by the Surveillance Epidemiology and End Results (SEER) of the National Cancer Institute for the years 1996—2004.
Surgical management provides the best potential for cure in these patients. Improvements in neo-adjuvant therapeutic approaches with chemoradiotherapy have allowed more patients to be candidates for aggressive trimodality approaches involving surgery [1–3]. However, oesophagectomy is associated with significant morbidity and mortality [4,5]. Morbidity is based mainly on pulmonary complications, anastomotic leakage, wound infections, vocal-cord paralysis and chyle leaks [5–7]. Patients with oesophageal cancer may present with dysphagia and weight loss, and frequently their nutritional status further deteriorates during delivery of neo-adjuvant therapy . With oesophagectomy, they are exposed to the potential for further nutritional depletion during postoperative recovery [8,9].
This study investigates an alternative postoperative management pathway following oesophagectomy consisting of enteral feeding with delayed oral intake compared with the conventional approach of performing a swallowing study while in hospital to rule out an anastomotic leak with early resumption of oral intake. The two pathways were compared with respect to anastomotic leak rate, length of hospitalisation and rates of unplanned re-admission to hospital to assess feasibility and safety of delayed feeding and to determine potential superiority of one approach versus the other. One of the further aims of this study was to analyse the potential for reduction in the length of hospitalisation.
Patients were identified from a prospective surgical database maintained by the Division of General Thoracic Surgery at the Mayo Clinic, Rochester, MN. The Mayo College of Medicine Institutional Review Board approved this study. A total of 492 consecutive oesophagectomies were performed between 1 July 2004 and 31 August 2008, of which 406 oesophagectomies were based on diagnosis of oesophageal cancer or Barrett’s oesophagus with high-grade dysplasia. Eighty-six oesophagectomies were performed for benign oesophageal disease, most commonly for end-stage achalasia and non-salvageable caustic strictures among other indications. In 10 of these oesophagectomies, a graft other than a gastric conduit was used. Procedures with gastrointestinal discontinuity, with colon interposition or free jejunal grafts for oesophageal reconstruction were excluded from this study. A priori, 10 patients declined consent to research participation and have been excluded from this analysis. Demographic and medical data of the remaining 386 patients were collected from medical records and analysed retrospectively.
In our practice, two different postoperative algorithms have been adopted after oesophagectomy (Fig. 1). All patients and their families receive nutritional counselling by a certified nutritionist during their postoperative recovery. This includes teaching to allow for self-administration of tube feedings at home. Follow-up is conducted by telephone periodically and patients receive an outpatient evaluation by their surgeon 4—6 weeks after their oesophagectomy.
The conventional approach comprises a radiographic contrast swallow study at 5—7 days after oesophagectomy and if no anastomotic leak is detected, patients proceed with gradually increasing oral intake, advancing to a soft post-gastrectomy diet prior to discharge home. A jejunostomy tube is occasionally used in case supplemental nocturnal tube feeding may be required. The alternative approach waives the routine postoperative radiographic contrast swallow examination, and all patients have a feeding jejunostomy tube placed at the time of oesophagectomy with the intention of exclusive enteral tube feeding over a period of 4 weeks before oral intake is gradually started. This approach has been increasingly implemented at our institution since July 2004. As the study design compared different practices within our department, the patients were assigned to each pathway group depending on the referral surgeon performing the case.
Data are presented as median and range for continuous variables, and as number and percentage for discrete variables. Comparison of patient characteristics and postoperative management group status was made using logistic regression, as well as to assess association between patient characteristics and anastomotic leak rate. A p-value less than 0.05 was used as the threshold for statistical significance. In addition, odds ratio estimates with 95% confidence intervals (CIs) are provided. Length of hospital stay, among patients surviving to discharge, was assessed using a Mann—Whitney test. Kaplan—Meier survival was used to estimate the cumulative probability of unplanned re-admission, with the log-rank test being used to assess association with postoperative management group. SAS statistical software, version 9, was used for statistical analysis.
There were 319 men (82.6%) and 67 women (17.4%) in the study population. A total of 110 patients (28.5%) were assigned to the alternative pathway and the remaining 276 (71.5%) to the conventional pathway. The patients’ characteristics were not significantly different between these two groups, except for body mass index (BMI) and the percentage of patients receiving neo-adjuvant therapy being higher in the alternative pathway, as presented in Table 1.
Median age at surgery was 64 years (range, 32—89 years), and the median BMI was 27.0 kg m−2 (range, 16.2—50.5 kg m−2). A smoking history was available in 379 of 386 patients, of which 12 patients reported smokeless tobacco usage, 105 patients never smoked (27.7%), 186 were former smokers (49.1%) and 76 patients were current smokers or considered as such with a history of having smoked within 4 weeks of diagnosis (20.1%). Alcohol use was recorded for 351 of 386 patients: 120 patients consumed no alcohol (34.2%), 121 had rare or occasional alcohol intake (34.5%), 66 reported moderate regular (18.8%) and 18 regular heavy intake or alcoholism (5.1%). In 26 of 351 patients, past alcoholism was documented (7.4%).
Of the 386 oesophagectomies, 115 were trans-hiatal oesophagectomies (29.8%), in 33 patients the tri-incisional (McKeown) approach was used (8.6%), 232 patients underwent a conventional transthoracic (Ivor Lewis) oesophagectomy (60.1%) and in six patients a minimally invasive Ivor Lewis oesophagectomy was performed using a combined thoracoscopic and laparoscopic approach (1.5%). For statistical analysis, the conventional and minimally invasive Ivor Lewis oesophagectomies have been combined. The association between the type of oesophagectomy and the algorithm of postoperative management did not reach statistical significance ( p = 0.08). Patients undergoing a trans-hiatal oesophagectomy showed a higher likelihood of being managed in the alternative pathway than the patients undergoing a transthoracic oesophagectomy (OR = 1.7; 95% CI: 1.1—2.8). Conversely, patients undergoing a tri-incisional oesophagectomy had a lower likelihood of being managed in the alternative pathway than transthoracic oesophagectomy patients (OR = 0.95; 95% CI: 0.4—2.2).
A nasogastric tube was inserted in 385 of 386 patients at the time of surgery. There were 311 patients who received a feeding jejunostomy intra-operatively (80.6%), while 75 patients were managed with a nasogastric tube only. Median length of operation was 323 min (range, 128—784 min). There was no intra-operative mortality.
Of the 386 patients, 370 (95.9%) underwent oesophagectomy for invasive oesophageal cancer, of which 318 (86.0%) were diagnosed with adenocarcinoma and 46 (12.4%) with squamous cell carcinoma. In six patients, other neoplasm types were diagnosed (1.6%). Of the 370 patients, 225 underwent neo-adjuvant therapy prior to surgery (60.8%). The mean largest tumour diameter at resection was 17 mm (range, 0—130 mm). Oesophagectomy for Barrett’s oesophagus with high-grade dysplasia was performed in 16 patients (4.1%).
There were 165 patients (42.8%) who suffered one or more postoperative complications, including atrial fibrillation in 45 (11.7%) and other arrhythmias in 19 patients (4.9%), wound complications in 37 patients (9.6%), anastomotic leaks in 36 patients (9.3%), respiratory insufficiency with need for prolonged ventilatory support in 36 patients (9.3%), pneumonia in 28 patients (7.3%), chyle leak in 19 patients (4.9%), recurrent laryngeal nerve injury in 15 patients (3.9%), pleural effusion in 13 patients (3.4%), delirium in nine patients (2.3%), gastric conduit necrosis in seven patients (1.8%) and sepsis in three patients (0.8%). Some patients suffered more than one complication, which is reflected by the cumulative percentage of greater than 100%. In 32 of 386 patients (8.3%), re-operation was required for various indications, including repair of anastomotic leakage, gastric conduit necrosis, small bowel obstruction unrelated to the feeding jejunostomy, wound dehiscence, chyle leak and other miscellaneous reasons. The overall in-hospital mortality was 3.6% (14 of 386). Of the 14 patients, nine died due to respiratory failure, two suffered cardiac arrest, one died during emergent re-operation for postoperative haemorrhage after initial oesophagectomy, one patient succumbed to a large anterior cerebral artery (ACA) territory infarct and one died after a complicated hospital course due to a necrotic gastric conduit. Of the 386 patients, four died within 30 days after discharge from the hospital (1.0%). Operative mortality, defined as in-hospital mortality or mortality within 30 days of operation, was therefore 18 out of 386 (4.6%).
A swallow study was conducted in 281 of 372 (in-hospital deaths excluded) patients at median postoperative day 6 (range, 2—51 days). Of note, five patients (4.5%) in the alternative pathway received a swallow examination without underlying clinical suspicion of an anastomotic leak for reasons such as evaluation of swallowing physiology with speech pathology. Techniques included aqueous contrast with gastrografin in 120 patients (31.1%), aqueous contrast followed by barium in 105 patients (27.2%), barium contrast in three patients (0.8%), grape juice swallow evaluation in 44 patients (11.4%) and a fluoroscopic video swallow study with occupational therapy in nine patients (2.3%). Of all patients undergoing a radiographic contrast swallow examination (237 of 281), 207 patients (87.3%) had an unremarkable report. In 11 of 237 (4.6%) patients, the radiographic swallow study detected an anastomotic leak. The study had a false-negative rate of 5.8%, as 12 of the 207 patients, with an unremarkable swallow study, had clinical signs of an anastomotic leak and were managed appropriately. Aspiration occurred in 14 of 237 patients (5.9%) during the swallow study. In one patient, the study had to be abandoned due to physical inability to swallow.
The hospital course of 28 patients was complicated by pneumonia, which was documented in the medical charts of 13 of 28 patients as aspiration pneumonia. In the conventional pathway group, three patients with an unremarkable radiographic contrast swallow study developed pneumonia related to aspiration, and six patients with observed aspiration during the contrast swallow study subsequently developed pneumonia.
In addition, two patients in the conventional pathway group developed aspiration pneumonia prior to undergoing a contrast swallow study. In general, two patients in the alternative pathway group developed aspiration, which was complicated by pneumonia. In addition, three of the five patients in the alternative pathway group who received a radiographic contrast swallow study showed signs of aspiration during the study. None led to any associated complications. No statistically significant differences were observed between the two groups for these events.
In 218 of 311 patients, the feeding tube was removed during a follow-up visit at the Mayo Clinic. The remaining patients were followed at institutions closer to home and therefore had the feeding tube removed at an outside institution. The feeding tube was removed after a median of 52 days (range, 4—434 days). In nine of these patients, the jejunostomy tube fell out accidentally, and it was not replaced. Complications related to the jejunostomy after discharge from the hospital occurred in eight of the 244 patients (3.3%), seven in the alternative pathway and one in the conventional pathway. These included erythema or dermatitis around the jejunostomy site and granulation tissue around the jejunostomy site in two patients. One patient developed a persistent jejunostomy site fistula requiring prolonged treatment, and one patient was diagnosed with an asymptomatic intussusception at the jejunostomy site.
Anastomotic leakage rate, hospital length of stay, as well as complications leading to re-admission have been analysed separately for the two different postoperative management approaches to identify potential outcome differences between groups.
Anastomotic leaks were detected in the conventional pathway group in 33 patients (12.0%), whereas patients in the alternative pathway group had only three anastomotic leaks (2.7%) ( p = 0.01) (Table 2). The type of resection was also significantly associated with the risk of an anastomotic leak ( p < 0.001): the leak rate for conventional or minimally invasive Ivor Lewis oesophagectomy was 3.8%, whereas the leak rate for trans-hiatal oesophagectomy was 17.4% and 21.2% for McKeown oesophagectomy. Relative to a patient with an Ivor Lewis oesophagectomy, a patient with a trans-hiatal oesophagectomy had an increased odd for an anastomotic leak (OR = 5.4, 95% CI: 2.35—12.19), as did a patient having a McKeown oesophagectomy (OR = 6.8, 95% CI: 2.35—12.19) (Table 2). This analysis excluded in-hospital deaths, although results were consistent when including inhospital deaths. Underlying comorbidities and neo-adjuvant therapy were not significantly associated with risk of an anastomotic leak.
Patients treated in the conventional pathway were discharged after a median of 9 days (range, 4—107 days), while in the alternative pathway discharge was significantly earlier with a median of 7 days (range, 5—54 days) ( p < 0.001). COPD ( p = 0.03) and underlying cardiovascular disease ( p = 0.009) prolonged the median length of stay by 3 days and by 1 day, respectively, than patients who did not have these comorbidities. Patients aged 65 years and older also had longer median hospital stays by 1 day ( p = 0.02) as compared with patients less than 65 years of age (Table 3). A multivariate analysis for factors affecting the length of hospitalisation included those factors found to be significant by univariate analysis. Use of the alternative pathway ( p < 0.001) and the absence of COPD as a patient comorbidity ( p = 0.03) remained statistically significant factors influencing shorter hospitalisation. Age of 65 years and greater and underlying cardiovascular disease, although significant on univariate analysis, were not found to be independent risk factors for a prolonged hospital stay by multivariate analysis ( p = 0.12 and p = 0.06, respectively).
There was no significant association between type of oesophagectomy and length of hospital stay (p = 0.13). Patients were discharged following trans-hiatal oesophagectomy after a median of 8.5 days (range, 4—107 days), following a transthoracic (Ivor Lewis) oesophagectomy with a median of 8 days (range, 5—73 days) and following a tri-incisional (McKeown) oesophagectomy with a median of 11 days (range, 5—65 days).
In general, information related to unplanned re-admission was available in 367 of 386 patients (95.1%). There were 103 patients who were re-admitted to the hospital after initial discharge, ranging from 1 to 135 days. In total, the cumulative probability of re-admission at 30 and 90 days was 20.9% (95% CI: 16.6—24.9) and 27.8% (95% CI: 23.0—32.3), respectively. The cumulative probability of readmission was not significantly different between the two pathways (p = 0.91). Twenty-eight patients in the alternative pathway group required re-admission, with cumulative probabilities of re-admission at 30 and 90 days of 22.4% (95% CI: 13.9—30.1) and 27.6% (95% CI: 18.4—35.9), compared with 75 re-admissions in the conventional pathway group, with cumulative probabilities of re-admission at 30 and 90 days of 20.3% (95% CI: 15.3—25.0) and 27.9% (95% CI: 22.2—33.2), respectively (Fig. 2). The main reasons for readmission included dehydration and/or emesis/nausea in 22 patients (21.4%), shortness of breath due to pleural effusion in 17 patients (16.5%), anastomotic leak or fistula in 12 patients (11.7%), dysfunction of the jejunostomy tube in 11 patients (10.7%), early stricture in 11 patients (10.7%), pneumonia due to aspiration in 10 patients (9.7%), wound infection in nine patients (8.7%), infectious pneumonia in seven patients (6.8%), deep vein thrombosis or pulmonary embolism in five patients (4.9%), sepsis in three patients (2.9%) and myocardial infarction in two patients (1.9%). Thirty-two patients had more than one reason for readmission. No statistically significant difference for specific complication rates was observed between the two postoperative management pathways.
After many years of managing post-oesophagectomy patients with a conventional approach consisting of routine evaluation of the oesophagogastric anastomosis with a radiographic contrast swallow study followed, if negative for leak, with early initiation of oral feeding, a segment of our thoracic surgical group explored an alternative postoperative care pathway for these patients. The alternative pathway, consisting of routine use of an intraoperatively placed feeding jejunostomy, delayed initiation of oral feeding and highly selective and therefore limited use of contrast swallow studies, was chosen. The adoption of this alternative pathway followed in-depth discussions with Dr Thomas W. Rice, who had begun using these measures in his practice in Cleveland, Ohio (T.W. Rice, personal communication, 2003). The chief perceived potential advantage of this alternative management strategy was the reduction in postoperative pulmonary complications due to aspiration secondary to either contrast swallow studies or overly ambitious early advancement of oral feeding.
Although we did not find a statistical difference in overall postoperative pulmonary complications, we did observe 11 documented cases of aspiration pneumonia in the conventional pathway (4.0%) and two in the alternative pathway group (1.8%) ( p = 0.30). These rates, for both the conventional and alternative pathway groups, compare favourably with current large-scale reports from other institutions [10–12], and with our previous reports [3,13]. We believe that a by-product of our oesophageal quality initiative was the simple recognition of aspiration pneumonia as a distinct negative outcome with clinically important ramifications for patients following oesophagectomy. This enhanced awareness by all members of the postoperative care team may have had, in and of itself, the beneficial effect of encouraging particular clinical approaches such as less aggressive advancement of oral feeding in the conventional group that may have the potential to decrease the risk of postoperative aspiration. This is also reflected by a reduction in the incidence of pneumonia after oesophagectomy since introducing the alternative pathway in our practice. In the current study, 7.3% of patients developed pneumonia during their postoperative recovery, while earlier studies at our institution reported rates of 12.3% for the years 1992—1995 and 14.2% for the years 1998—2003, respectively [3,13].
In our cohort, the alternative postoperative management strategy was associated with a significantly shorter length of hospital stay (median = 7 days) without a higher rate of readmission or postoperative complications. This is despite having a significantly higher proportion of patients in the alternative pathway group that had prior neo-adjuvant chemotherapy and radiotherapy. Furthermore, these patients did not appear to have any additional complications after dismissal. In fact, reasons and rates for unplanned readmission to the hospital were not significantly different between both postoperative management groups.
In our cohort of conventional postoperative management, a median length of hospitalisation of 9 days was observed as compared with other large single-centre studies at high-volume institutions which report a median length of hospitalisation of 11—14 days [13,14].
Prior to early oral intake, patients in our conventional management pathway undergo a radiographic contrast swallow. A routine contrast swallow study for evaluation of postoperative anastomotic leaks following oesophagectomy has been previously reported to be a diagnostic test with high specificity but low sensitivity, especially if water-soluble contrast is used [15,16]. Although the current investigation did not aim specifically to evaluate the sensitivity and specificity of routine radiological examinations for postoperative anastomotic leaks after oesophagectomy, we were able to report that the overall false-negative anastomotic leak rate was 5.8%. Other studies report false-negative rates of 5.4% and higher for the Gastrografin swallow test [16,17].
In comparing patients with early oral intake (conventional pathway) with those with delayed institution of oral feeding (alternative pathway), there was a significantly lower anastomotic leak rate detected in the latter group. As the majority of leaks were discovered by radiographic swallow examination, it remains possible that in the alternative pathway group with delayed feeding, a number of subclinical leaks were present. Similarly, early feeding may indeed increase the risk of anastomotic failure, by early distension of the gastric conduit leading to either compromised blood flow by limiting venous drainage or increased tension on the anastomosis.
Health-related quality of life is an important outcome measure in addition to long-term survival rates, mortality and complication rates of extensive oncological surgery, such as oesophagectomy . Various studies to date have shown that overall quality of life seems not be impaired in long-term survivors following oesophagectomy [19–24]. There are no studies to date, however, evaluating patient’s quality of life with respect to early feeding after oesophagectomy versus delayed oral intake with enteral feeding via jejunostomy.
In summary, an alternative postoperative care strategy for patients following oesophagectomy which delays the initiation of oral intake and avoids early contrast swallow studies is associated with a lower detected anastomotic leak rate and shortened length of hospitalisation, without leading to a higher risk of complications or re-admission to hospital. This alternative pathway is therefore both safe and feasible for patients following oesophagectomy.
The authors would like to sincerely acknowledge Dr Thomas W. Rice as the initial proponent of an alternative post-oesophagectomy care pathway as described in this study. We appreciate the collegial dialogue he participated in with us throughout the implementation of this project.
Presented at the 17th European Conference on General Thoracic Surgery, Krakow, Poland, May 31—June 3, 2009.
Supported by funds from Mayo Foundation.