Early Readmissions after Esophagectomy for Esophageal Adenocarcinoma: Does Facility Case-Volume Matter?

Increased esophagectomy procedures over the past four decades have correlated with the rise in incidence of esophageal adenocarcinoma. Despite advances in technology and procedural expertise, esophagectomy remains a high-risk surgical procedure. Higher volume facilities have more experience with esophagectomy and would be expected to have a lower incidence of surgical complications and attendant morbidity and mortality. By analyzing information from a nationwide United States hospital database, we sought to find out if there is a significant difference between facilities stratified by case volume, with regards to 30-day readmission after esophagectomy. The findings of this study indicated that even with a large applied differential, early readmissions did not differ significantly between high- and low-volume centers. Also, analyzed and discussed were any associated demographic and comorbidity factors as they relate to early readmissions after esophagectomy for esophageal adenocarcinoma across the country. This is the first study to specifically address these variables.


Introduction
e incidence of esophageal adenocarcinoma in the United States has increased dramatically since the 1970s and has been attributed to a contemporaneous rise in the prevalence of obesity, Barrett's esophagus and gastroesophageal reflux disease [1]. Concurrently, there has been an increased number of esophagectomies performed for esophageal cancer. An analysis of United States hospital data from 2001 to 2010 demonstrated an overall increased trend in esophagectomy procedures performed for this indication with a concomitant reduction in the mortality rate [2].
Esophagectomy for esophageal cancer remains a highrisk procedure with a substantial morbidity and mortality rate [3]. Previous studies including meta-analysis suggest that esophagectomy performed at low-volume hospitals was associated with a significant increase in incidence of inhospital and 30-day mortality [4]. More recently, with hospital case volume assigned to a threshold of 10 cases per year, there was no significant difference noted in risk-adjusted outcomes between high-volume centers (HVCs) and low-volume centers (LVCs) [2].
Hospital readmissions following esophagectomy range from 12-25% and are associated with decreased patient survival [3,5,6]. However, data regarding risk factors for early (30-day) readmission after esophagectomy have been conflicting. Also, it is not known whether the incidence of early readmission following esophagectomy is related to the annual case volume of the performing center.
We utilized a nationwide readmission database to study 30-day readmissions following esophagectomy performed for esophageal cancer. e aim of this study was two-fold; first, to compare the incidence of 30-day readmissions among centers stratified by annual case volume. Second, to study patient demographics, comorbidities, and associated diagnosis and procedure codes as they relate to early readmissions following an esophagectomy.

Data Source.
We obtained data for this study from the 2013 Nationwide Readmissions Database (NRD), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality (AHRQ), which is designed to provide information on national readmission rates in the U.S. e NRD is an annual database and contains data from community, specialty, and public hospitals and academic medical centers including pediatric hospitals. e NRD is a "discharge-level" file, meaning that each record represents a single discharge corresponding to an inpatient stay. Should the patient have multiple hospital visits in a particular year, the NRD will include separate records corresponding to each inpatient stay. e NRD includes discharges on patients aged 1 year and older. Each discharge entry includes patient demographic details, 1 primary discharge diagnosis (based on the International Classification of Diseases, Ninth Revision, and Clinical Modification, ICD-9-CM, diagnosis codes), 1 to 24 ICD-9-CM secondary diagnoses, 1 to 15 ICD-9-CM procedural codes, hospitalization information, and hospital data. Discharge weights are provided to generate national estimates; it estimates roughly 36 million discharges in the United States. All data in this study are reported as national level estimates. e study excluded any protected health information and was deemed exempt from institutional board review.

Esophagectomy Definition, Study Demographics, and
Readmissions. For the purpose of this study, we included all patients 18 years of age and older who underwent esophagectomy, total or partial, performed for the indication of esophageal cancer during the period January 1-December 31, 2013 and were discharged alive following surgery. We used previously described International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic and procedural codes [2]. ese included ICD-9-CM diagnosis codes for esophageal cancer 150.0-150.5, 150.8, and 150.9 and ICD-9-CM procedure codes for partial esophagectomy 42.41, 42.5, 42.51, and 42.52 and for total esophagectomy 42.42 and 42.11. e total number of esophageal resections was calculated for each hospital. Hospitals were then classified as either HVCs (10 or more cases during the study period) or LVCs (fewer than 10 cases during study period). For each esophagectomy, it was determined if the patient had a subsequent readmission in a 30-day period.
Patient demographics, diagnoses, procedures, and hospitalization data were obtained for the extracted cases. e occurrence of postoperative complications (POC) following esophagectomy [2], the incidence of patient comorbidities using the Elixhauser scale [7], and the requirement for intensive care unit (ICU) stay [8] were assessed for using previously described methods.
We calculated hospital length of stay (LOS) and hospitalization costs (obtained by multiplying total charges by cost-to-charge ratios and rounded to the nearest $1,000). We also determined the most frequently associated diagnostic and procedural codes associated with early readmissions. Specifically, in patients with a 30-day readmission after esophagectomy, we extracted all ICD-9-CM primary diagnoses codes that occurred with a frequency of 2% or greater. Similarly, we also extracted all ICD-9-CM procedural codes that occurred at a frequency of 3% or greater.

Statistical Analysis.
Statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina, USA). e Kolmogorov-Smirnov test was used to determine the normality of the data elements under study. e Kruskal-Wallis and chi-square test were used to compare differences in continuous and categorical variables, respectively. Univariate and multivariate logistic regression analyses were performed with the presence or absence of 30day readmission as the dichotomous outcome variable. Covariates that were tested for association with readmission included variables that were identified as significant at the univariate level. We tested all between-variable estimated correlation coefficients and determined that multicollinearity was not a problem. Odds ratios (ORs), adjusted odds ratios (aORs), and 95% confidence intervals (CIs) were reported to identify the strength and significance of readmission and other covariates on the likelihood of an association. e threshold for significance for all analyses was p < 0.01.

Results
In 2013, there were a total 2,017 esophagectomies performed for indication of esophageal cancer meeting our study criteria.
Comparing patients who did not undergo a 30-day readmission following esophagectomy to those who did, the major demographic characteristics, illness severity, and hospital course are outlined in Table 4.
For patients undergoing a 30-day readmission, the major associated comorbidities and postoperative complications are outlined in Table 5.

Discussion
Readmission after surgery continues to be a significant healthcare problem, and recent policy changes that include substantial financial penalties have made readmission in a Table 1: Comparison of esophagectomy performed at high-and low-volume centers for esophageal cancer in patients 18 years of age and older who survived their initial hospital stay. High-volume centers were defined as performing 10 or more cases per year, while low-volume centers performed 9 or less cases annually. IQR interquartile range.

Variable
High-volume center   crucial pay-for-performance program in the U.S [9]. Our results indicate that the incidence of 30-day readmission after an esophagectomy continues to be high at 17.20% and is consistent with previous finds [5,6]. e novel aspect of this study is that it compares the incidence of early readmissions following esophagectomy among centers stratified by annual case volume. Patient characteristics at both the centers did not differ significantly in regard to median age, comorbidity score, and the median time to readmission or costs. Importantly, the incidence of 30-day readmissions did not significantly differ between the two types of centers. Even when we reassigned the case threshold for HVCs to ≥30 cases per year, the difference did not achieve statistical significance (data not shown). While HVCs did demonstrate an advantage in shorter hospital stays and a lower incidence of POCs, these have been previously shown to be nonsignificant in multivariate analysis [2]. While the definition of high-and low-volume centers in the literature is extremely variable, these data question the perceived benefits of esophagectomy performed at a HVC. We analyzed the diagnoses and procedures associated with 30-day readmission following an esophagectomy. An infectious etiology was noted as the primary cause for readmission in 17.5% of cases including septicemia (6.1%), pneumonia (5.1%), and empyema (2.6%). Postoperative pulmonary complications are recognized as the most frequent systemic complications after esophagectomy, and intensive preoperative respiratory rehabilitation has been demonstrated to decrease these complications [10]. Esophagogastroduodenoscopy (EGD) was noted to be the most frequently performed procedure in 19.3% of patients following a 30-day readmission after esophagectomy.
Following an esophagectomy, up to 10% of patients developed an anastomotic leak and up to 22% developed an anastomotic stricture; 34% of these strictures are manifested within 1-3 months [11]. In this context, we noted that 6.3% of our readmitted cohort underwent esophageal stenting and 3.4% underwent esophageal dilation. e overall healthcare burden associated with 30-day readmissions after esophagectomy was substantial. is was realized in the significantly high hospital-related stays and costs as well as unfavorable outcomes including ICU stay and death. Mortality and ICU stay were strongly associated with an infectious, cardiopulmonary, or renal complication. Factors associated with early readmission included lengthier hospital stays and higher hospital costs following the index surgery as well as a greater incidence of comorbidities and POCs. All of these likely represent a more complicated initial hospital course and consequently portend early readmission. erefore, the subset of patients displaying these high-risk characteristics would possibly benefit from more detailed evaluation before discharge.
On multivariate analysis, we observed that cardiovascular and renal complications present at the time of esophagectomy were independently associated a higher incidence of subsequent 30-day readmission. A higher preoperative comorbidity score has been inconsistently reported as a risk factor for readmission. For example, Fernandez et al. reported that a Charlson comorbidity index of 3+ was strongly associated with readmission [5]. However, a subsequent study reported that the only significant preoperative predictor of readmission was delivery of induction therapy [6]. Similarly, the incidence of POCs has been reported to be significantly associated with an increased   [12]. However, we were unable to find evidence for this in multivariate analyses. ere are limitations to this study; several of these have previously been elaborated upon in an analysis of the related Nationwide Inpatient Sample (NIS) database [2]. e HCUP family of databases including the NIS and NRD do not contain data pertaining to tumor type and stage, pulmonary function, performance status, or the use of neoadjuvant therapy. e study relies solely on the use of ICD-9-CM codes for case and procedure identification. ere is a lack of information on the surgeons' specialty or case volume, which may be more closely related to outcomes than hospital volume [13]. Specific to the NRD is a lack of information on patient race, the limitation of data to a single calendar year, and the possibility that patients were readmitted to a hospital in a different state, whereupon they would lose their initial identification code in the database. However, the NRD does have an advantage over the frequently queried Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset, which excludes patients <65 years of age [6].

Conclusions
Readmissions within a 30-day period are common after esophagectomy performed for esophageal cancer. Even with a large applied differential, early readmissions did not differ significantly between high-and low-volume centers. Patients with a complicated medical course after esophagectomy, greater comorbidities, and POCs demonstrate a higher incidence of readmission; cardiovascular and renal comorbidities are independently associated with an increased risk. Infectious etiologies are the leading cause of early readmission, while EGD with stent placement and dilation is the most frequently performed medical procedure. Based on patient characteristics and outcomes during initial hospital stay, patients at high risk for readmission may be identified and should be evaluated carefully before clearing for hospital discharge.

Data Availability
e data used to support the findings of this study are available from 2013 Nationwide Readmissions Database (NRD) of the Healthcare Cost and Utilization Project (HCUP), through the United States Agency for Healthcare Research and Quality (AHRQ).