Emergency surgery for gastrointestinal cancer: A nationwide study in Japan based on the National Clinical Database

Abstract Background Emergency gastrointestinal surgery, although rare, is known for its high mortality and morbidity. However, the risks of emergency surgery for gastrointestinal cancer have not been investigated in depth. This study aimed to investigate the impact of emergency surgery on mortality and morbidity in patients with gastrointestinal cancers and to identify associated risk factors. Methods We extracted data from the National Clinical Database, a nationwide surgery registration system in Japan, for patients with gastrointestinal cancer who underwent esophageal resection, total gastrectomy, distal gastrectomy, right hemicolectomy, or low anterior resection between 2012 and 2017. The impacts of emergency surgery on 30‐day mortality and incidence of overall postoperative complications were compared with those of non‐emergency surgery. Risk factors for mortality and overall postoperative complications were then sought in patients who underwent emergency surgery. Results Thirty‐day mortality and incidence of overall postoperative complications were significantly higher in emergency surgeries for gastric, colon, and rectal cancers than in non‐emergency surgeries (odds ratios 4.86‐6.98 and 1.68‐2.18, respectively; all P < .001). Various risk factors were identified in the group that underwent emergency surgery, including preoperative sepsis and lower body mass index. Some of the risk factors were common to all types of surgery and others were specific to a certain type of surgery. Conclusion The actual risk of emergency surgery and the risk factors for overall postoperative complications in emergency cases are shown to serve as a reference for postoperative management. Emergency surgery had an additional burden on patients depending on the type of surgery.


| INTRODUC TI ON
Emergency surgery for gastrointestinal cancer is known to have high mortality and a high incidence of postoperative complications. 1 Postoperative mortality is a worldwide problem and was reported to be the third leading cause of death globally in 2016. 2 Emergency gastrointestinal surgery is usually performed only when patients have critical sudden-onset symptoms, such as intestinal bleeding, intestinal stenosis, and peritonitis as a result of gastrointestinal perforation. [3][4][5][6][7][8][9][10][11][12][13] Therefore, patients who require emergency surgery are in a worse condition than those who undergo non-emergency surgery. However, with the exception of some procedures, such as stent placement for colonic stenosis, emergency surgery is likely to be unavoidable despite the poor condition of these patients. 14 Although it is well known that patients undergoing emergency surgery have higher mortality and morbidity than those undergoing non-emergency surgery, the associated risk factors in emergency gastrointestinal surgery remain unclear because emergency gastrointestinal surgery is rare and most of the previous studies have included small numbers of patients. [3][4][5][6][7][8][9][10][11][12][13] The aims of this study were to investigate the impact of emergency surgery for gastrointestinal cancer on mortality and morbidity using the largest patient registry in Japan, to identify the risk factors for 30-day mortality and incidence of overall postoperative complications in these patients who undergo emergency surgery. 15

| Study design and setting
This retrospective observational study used data from the National Clinical Database (NCD), which is a nationwide surgical registration system in Japan that contains data on early clinical outcomes, including postoperative mortality and intraoperative and postoperative complications. The NCD is linked to the board certification system for gastrointestinal surgery and covers more than 95% of surgical cases in Japan. 15,16 Detailed data are collected for five types of major gastrointestinal surgery, and risk models for mortality and postoperative complications have been created in an effort to achieve better postoperative outcomes. [17][18][19][20][21] We extracted the data for patients who underwent esophageal resection, total gastrectomy, distal gastrectomy, right hemicolectomy, or low anterior resection from 2012 to 2017 from this database. Data for patients aged ≥18 years who had a malignant tumor at the site of resection were included, and data for those with missing values for potential risk factors or outcome variables and those with an abnormal value for length of hospital stay (negative or zero) or body mass index (BMI) (<10 or ≥200) were excluded. Length of stay was defined as the time from the date of admission to the date of discharge. The study was approved by the Ethics Committee of Kyoto University.

| Actual risk of emergency surgery for gastrointestinal cancer
We investigated the actual risk associated with each type of surgery by comparing the emergency surgery and non-emergency surgery.
Primary outcomes were 30-day mortality and incidence of overall postoperative complications. We also set intraoperative outcomes, postoperative outcomes, and the incidence of each postoperative complication as secondary outcomes (Appendix S1).

| Risk factors for 30-day mortality and incidence of overall postoperative complications after emergency surgery
We focused on the patients who underwent emergency surgery and investigated the risk factors for the primary outcomes of the first analysis, namely, 30-day mortality and incidence of overall postoperative complications for each type of emergency surgery. We analyzed both preoperative and intraoperative risk factors (Appendix S2).

| Statistical analysis
Continuous variables are shown as the median and interquartile range and categorical variables are shown as the number and percentile. The risk of emergency surgery compared with non-emergency surgery was calculated using a univariable logistic regression model and is reported as a crude odds ratio (OR) with a 95% confidence interval (CI) to demonstrate the actual risk of emergency surgery. The risk factors for 30-day mortality and overall postoperative complications after emergency surgery were investigated first using univariable logistic regression model and then using a multivariable logistic regression model to identify independent risk factors.
All factors included in the univariable analysis were considered to be clinically important and included as covariates in the multivariable analysis to identify true risk factors in the emergency surgery group. All P-values were two-sided and P-values less than 0.05 were considered statistically significant. The statistical analyses were performed using R software (version 3.5.0, 2018; R Foundation for Statistical Computing, Vienna, Austria).

| Patient characteristics
A total of 32 425 patients who underwent esophageal resection for esophageal cancer (non-emergency surgery, n = 32 315, 99.7%; emergency surgery, n = 110, 0.3%) were included in the study (Figure 1). The preoperative characteristics in the emergency surgery group included high proportions of patients with dyspnea, dependence in activities of daily living (ADL), metastatic cancer in another organ, weight loss, blood clotting defects, and sepsis. Their intraoperative characteristics included less use of thoracoscopy, high American Society of Anesthesiologists physical status (ASA-PS), and advanced T stage (Table 1).

| Actual risk of emergency surgery
There were no deaths and only a small number of postoperative complications in the emergency surgery group (Table S1). Therefore, we did not calculate the ORs for emergency esophageal resection in comparison with non-emergency surgery.

| Patient characteristics
A total of 95 934 patients underwent total gastrectomy for gastric cancer ( Figure 1). Non-emergency surgery was performed in 94 959 patients (99.0%) and emergency surgery was performed for 975 patients (1.0%). The preoperative characteristics in the emergency surgery group included high proportions of patients with low BMI, dyspnea, dependence in ADL, ascites, metastatic cancer in another organ, weight loss, blood clotting defects, and sepsis. Preoperative blood transfusion and chemotherapy were more common in the emergency surgery group. Intraoperative characteristics included less use of laparoscopy, high ASA-PS, advanced TNM stages, and a high frequency of residual tumor (Table 1).

| Actual risk of emergency surgery
The primary outcome of 30-day mortality was significantly higher in the emergency surgery group than in the non-emergency surgery group (OR: 6.12, 95% CI: 4.50-8.32, P < .001). The incidence of overall postoperative complications was also significantly higher in the emergency surgery group (OR: 1.68, 95% CI: 1.48-1.91, P < .001) ( Table 2).
For the secondary outcomes, incidence of intraoperative and postoperative adverse events was significantly higher in the emergency surgery group (Table 3). The frequencies of intraoperative and postoperative blood transfusion and mechanical ventilation and      Table 3). The number of patients with each complication is listed in Table S1.

| Actual risk of emergency surgery
The 30-day mortality was significantly higher in the emergency surgery group than in the non-emergency surgery group (OR: 8.63, 95% CI: 6.65-11.18, P < .001). The incidence of overall postoperative complications was also significantly higher in the emergency surgery group (OR: 1.78, 95% CI: 1.61-1.98, P < .001) ( Table 2).
For the secondary outcomes, the incidence of intraoperative and postoperative adverse events was higher in the emergency surgery group (  Table 3). The numbers of patients with each complication are shown in Table S1.
Compared with the non-emergency surgery group, the emergency surgery group had significantly shorter anesthesia and operating times (312 vs 265 minutes, P < .001 and 255 vs 208 minutes, P < .001, respectively), significantly greater estimated blood loss (110 vs 215 mL, P < .001), and a significantly longer hospital stay (18 vs 22 days, P < .001; Table 4).

| Patient characteristics
The data for the 114 068 patients in the database who underwent right hemicolectomy for colon cancer were included in the analysis; 109 169 (95.7%) of these patients underwent non-emergency surgery and 4899 (4.3%) underwent emergency surgery ( Figure 1).
Analysis of the preoperative characteristics in the emergency surgery group revealed a high frequency of low BMI, dyspnea, dependence in ADL, ascites, metastatic cancer in another organ, weight loss, blood clotting defects, and sepsis. Intraoperative characteristics included less use of laparoscopy, a high frequency of concurrent surgery, poor ASA-PS, advanced TNM stages, and a high frequency of residual tumor (Table 1).

| Actual risk of emergency surgery
The 30-day mortality was significantly higher in the emergency surgery group than in the non-emergency surgery group (OR: 6.98, 95% CI: 5.87-8.30, P < .001). The incidence of overall postoperative complications was also significantly higher in the emergency surgery group (OR: 2.18, 95% CI: 2.05-2.31, P < .001) ( Table 2).
The incidence of many intraoperative and postoperative adverse events was higher in the emergency surgery group than in the non-emergency surgery group ( intraoperative and postoperative blood transfusion, unscheduled intratracheal intubation, and mechanical ventilation and the incidence of deep incisional SSI, wound disruption, pneumonia, renal dysfunction, prolonged disturbance of consciousness, cardiac arrest, and sepsis was particularly high in the emergency surgery group (OR > 3; Table 3). The numbers of patients with each complication are shown in Table S1.
Compared with the non-emergency surgery group, the emergency surgery group had significantly shorter anesthesia time and operating times (256 minutes vs 220 minutes, P < .001 and 198 minutes vs 165 minutes, P < .001, respectively), significantly higher estimated blood loss (61 mL vs 160 mL, P < .001), and a significantly longer hospital stay (18 days vs 21 days, P < .001; Table 4).  (Table 1).

| Actual risk of emergency surgery
The 30-day mortality was significantly higher in the emergency surgery group than in the non-emergency surgery group (OR: 4.86, 95% CI: 2.98-7.93, P < .001). The overall incidence of postoperative complications was also significantly higher in the emergency surgery group (OR: 1.70, 95% CI: 1.51-1.93, P < .001) ( Table 2).
Intraoperative and postoperative adverse events were more frequent in the emergency surgery group than in the non-emergency surgery group (Table 3). The frequencies of intraoperative and postoperative blood transfusion and mechanical ventilation and the incidence of renal dysfunction, prolonged disturbance of consciousness, cardiac arrest, acute myocardial infarction, and sepsis was particularly high in the emergency surgery group (OR > 3; Table 3). The numbers of patients with each complication are listed in Table S1.  blood loss (75 mL vs 225 mL, P < .001), and a significantly longer hospital stay (20 days vs 25 days, P < .001; Table 4).

| Risk factors for 30-day mortality
We did not analyze the risk factors for 30-day mortality after esophageal resection because no patients in this group died. Moreover, the numbers of events were too small in the groups that underwent the other four types of emergency surgery to perform multivariable analyses; therefore, only univariable analysis was performed (Table S2). Several factors were identified as potentially critical (OR > 3) for 30-day mortality (Appendix 3).

| Risk factors for postoperative overall complications
The  (Table 5).

| D ISCUSS I ON
In this study, we investigated the frequencies of five types of major emergency surgery for gastrointestinal cancer (for which emergency surgeries accounted for 0.3%-4.3% of all surgeries) and calculated the actual 30-day mortality and incidence of complications associated with all of these operations except for esophageal resection.
Our results show that the risks of these four types of gastrointestinal surgeries were significantly higher when they were performed on an emergency basis. However, these findings do not imply that As in previous studies using NCD [17][18][19][20][21]   to the nutrition status than those undergoing another procedure in emergency cases. Dependence in ADL and COPD were associated with high risk in colorectal surgery, and a history of CVD was associated with high risk in gastrectomy. Patients who underwent concurrent surgery with distal gastrectomy or right hemicolectomy were also at high risk. Diverting stoma was not associated with the incidence of overall postoperative complications in low anterior resection. These results suggest that being in a poor state preoperatively was more critical for patients undergoing emergency surgery than cancer progression or the effects of preoperative chemotherapy. As in the above description about the incidence of postoperative complications, some risk factors were common across the four types of emergency surgery and others were different, suggesting that all emergency gastrointestinal surgeries involve a physical burden but that some have an additional burden depending on the type of surgery.
The main strength of this study is that it used nationwide data in Japan. The NCD database covers almost all surgeries performed in Japan because it is linked to the board certification system in Japan.
Moreover, many of the variables studied, including preoperative factors and postoperative complications, have been collected for five types of surgery. Emergency surgery is rare and only a large-scale database like the NCD can clarify its actual risks. Furthermore, only patients with cancer were included in this study, which eliminated the heterogeneity between benign and malignant diseases. We also included data for the five types of major surgery performed for gastrointestinal cancer to be able to provide gastrointestinal surgeons with useful information regarding the risks of these surgeries when performed on an emergency basis. However, the study also has some limitations, particularly its retrospective design and the potential for recall bias, and transcription errors. Also, data input was dependent on each institution. Nevertheless, we believe that the influence of theses biases was limited because the data were entered into the NCD on a yearly basis and it has been reported that there is little difference between the data on medical charts and those in the NCD. 28 In conclusion, emergency surgery for gastrointestinal cancer was associated with high 30-day mortality and morbidity in Japan.
Emergency surgery had an additional burden on patients depending on the type of surgery. The actual risk of emergency surgery and the risk factors for overall postoperative complications in emergency cases are shown to serve as a reference for postoperative management.