Ninety Days and In-Hospital Mortality after Gastrointestinal and Hepatopancreatic Biliary Surgery- A Case Series Analysis.

Aim: The aim was to do a retrospective analysis and audit of our 90 days and in-hospital mortality after gastrointestinal and hepatopancreatic biliary surgery performed in our department and analyze factors predicting it. All patients who underwent gastrointestinal and hepatopancreatic biliary surgery in our department in the last 3 years were evaluated for 90 days postoperative and in-hospital mortality and various factors affecting it. Categorical values were analyzed using the chi-square test or sher's exact test wherever appropriate. Continuous variables were analyzed using the student t-test for parametric data and Mann Whitney U test for nonparametric data after skewness and kurtosis analysis. Multivariate analysis was done using logistic regression analysis. A p-value less than 0.05 was considered statistically signicant. Statistical analysis was done using SPSS version 23(IBM). On multivariate and independently predicted 90 mortality.

Surgeons are sometimes not comfortable discussing their outcome data and there is an increasing need for transparency in reporting surgical outcomes. A surgical mortality audit is an important tool to improve quality. There are some reports [3] that mortality audits improve surgical performances and make surgeons understand the role of better communication with the patients regarding real risk-bene t ratios and stop them from giving unrealistic hope to patients and their relatives.
This analysis aims to do a mortality audit of our data about in-hospital and 90 days mortality. We also evaluated various factors responsible for mortality via univariate and multivariate analysis.

Patients And Methods:
All patients who underwent gastrointestinal and hepatopancreatic biliary surgery in our department in the last 3 years were evaluated for 90 days postoperative and in-hospital mortality and various factors affecting it. We evaluated various factors like age, sex, type of surgery, emergency surgery. CDC grade of a wound, American society of anesthesiology grade, blood product used, operative time, technical or nontechnical complications, acute kidney injury, all types of leaks, bile leaks, anastomotic leaks, open or laparoscopic surgery, overall morbidity, and malignant diseases for their association with 90 days and inhospital mortalities.

Statistical analysis:
Categorical values were analyzed using the chi-square test or sher's exact test wherever appropriate.
Continuous variables were analyzed using the student t-test for parametric data and Mann Whitney U test for nonparametric data after skewness and kurtosis analysis. Multivariate analysis was done using logistic regression analysis. A p-value less than 0.05 was considered statistically signi cant. Statistical analysis was done using SPSS version 23(IBM).

De nitions:
Mortality: In-hospital, mortality was de ned as any cause of mortality when a patient is admitted to the hospital. 90 days mortality was de ned as any cause of mortality in 90 days post-operative period.

Acute kidney injury:
Acute kidney injury is de ned according to acute kidney injury network classi cation [4,5]. Any grade of acute kidney injury was considered signi cant.

Intraoperative hypotension:
Intraoperative hypotension was de ned as systolic arterial pressure below 80 mmHg, a decrease in systolic arterial pressure by 20% below baseline, or vasopressors requirement. [6] Nontechnical and technical complications: Non-technical complications were de ned as perioperative complications related to patients' physiological health or comorbidities (e.g. acute kidney injury, ARDS, acute respiratory failure, cardiac complications, etc.), rather than to surgical procedures or techniques.
Technical complications were de ned as perioperative complications related to surgical procedures or techniques (e.g. bleeding, leaks, sepsis, etc.). ARDS was de ned according to the Berlin de nition [7]. Acute myocardial infarction and postoperative left ventricular dysfunction were diagnosed as per cardiologists' opinion based on cardiac markers, electrocardiogram, and echocardiography. Pulmonary embolism was con rmed using a contrast-enhanced CT scan.
Centre of Disease Control Grading: We also de ned surgical wounds according to the Centre for Disease Control as clean (grade 1), cleancontaminated (grade 2), contaminated (grade 3), and dirty (grade 4).

Major and Nonmajor surgery:
We de ned surgeries with literature proven negligible mortality rates like laparoscopic cholecystectomy, All hernia surgeries, laparoscopic appendicectomies as nonmajor surgeries, and other surgeries as major surgeries. All emergency surgeries were also de ned as major surgeries.
Results: 412 patients underwent gastrointestinal and hepatobiliary surgery in the last 3 years at our institute.

days and in-hospital mortality:
Ninety days all-cause mortality was 5.8%, all-cause in-hospital mortality was around 4.6%. 90 days mortality in elective and emergency surgeries were respectively 3.2% and 18%. In-hospital mortality in elective and emergency surgeries were respectively 2.35% and 15.2%. 222 surgeries were de ned as major surgeries and 190 as nonmajor (including 143 laparoscopic cholecystectomies). 90 days mortalities and in-hospital mortality in elective major surgeries were respectively 6.7% and 4.8%.
The type of surgery is mentioned in Table 1.  We also analyzed factors affecting survival via cox regression analysis. It also showed age, non-technical complications, and emergency surgery was associated with worse survival over 90 days. [ Table 3]. Figure 1 shows Kaplan Meier survival curved comparing 90 days survival in patients who developed nontechnical complications vs patients who did not develop them. Discussion: Perioperative mortality is one of the most important problems the surgical community must face.
Perioperative mortality ranges from 0.1% to as high as 27-30%, depending on the type of surgery. [8,9]. Gastrointestinal and hepatobiliary surgery is technically demanding procedures and has among the highest perioperative mortality rates. [10,11,12]. Another issue is surgeons rarely audit their mortality data scienti cally and in-process rarely know what factors might be responsible for poor outcomes in their patients.
This study aimed to do a mortality audit for patients operated for gastrointestinal and hepatobiliary surgeries in our department and study factors responsible for 90 days and in-hospital mortality.
Our elective 90 days and in-hospital mortality rates after elective surgeries were around 3.2% and 2.3% respectively and 18% and 15.2% respectively, which is comparable to what Sørensen et al. [12] described in their study. They showed 30 days mortality of 2.8% and 13% respectively in elective and emergency surgeries.
Our overall 90 days and in-hospital mortality rates were 5.6% and 4.2% which were comparable to published literature. [13].
We chose to study 90 days mortality instead of 30 days mortality as mayo et al [14] described 90 days mortality should be the standard criteria to describe perioperative mortality.
As we have described in results various factors were associated with 90 days and in-hospital mortality but on multivariate analysis nontechnical complications, age and emergency surgery independently predicted both 90 days and in-hospital mortality. Post-operative acute kidney injury was associated with in-hospital mortality independently however it was not associated with 90 days mortality after multivariate analysis which may imply that patients who recovered from post-operative acute kidney injury did well subsequently.
90 days mortality and in-hospital mortality in major surgeries in our series were 6.7% and 4.8% respectively. Major surgery was not independently associated with 90 days or in-hospital mortality after multivariate analysis. Heyer et al [15] in their recently published study consisting of a large cohort of complex gastrointestinal surgeries showed textbook outcomes are increased over time, they showed 90 days mortality of around 9.2% in a cohort of more than 31000 patients operated between 2014-2016. In our series complex, gastrointestinal surgery for malignancy showed similar results of 9.3% 90 days mortality and 5.3% in-hospital mortality. However, surgery for the malignant disease was not associated with 90 days or in-hospital mortality.
Type of surgery like liver resections or transplants, HPB surgery, upper gastrointestinal surgery, small intestinal surgeries, colorectal surgeries, hernia, and other surgeries were not associated with 90 days or in-hospital mortality. Colorectal surgery was not associated with 90 days mortality, it was associated with in-hospital mortality on univariate analysis but failed to show association on multivariate analysis. In our series, nontechnical complications were independently associated with 90 days and in-hospital mortality after multivariate logistic regression analysis and their strength of associated was very high with odds ratios of 61.15 and 94.23 respectively for 90 days and in-hospital mortality, suggesting the need to concentrate on preoperative, intraoperative, and postoperative critical care management to prevent and treat such complications. Pre and intraoperative factors like CDC grading of wound classi cation, American society of anesthesia score, number of blood transfusions, duration of surgery predicted 90 days, and in-hospital mortality on univariate analysis but failed to show independent association on multivariate analysis and they might have some role to play in the development of nontechnical complications.
Technical complications like intraoperative bleeding, anastomotic leaks, or bile leaks have no association with in-hospital mortality. Anastomotic leaks and bile leaks had an association with 90 days mortality on univariate analysis but had no independent association after multivariate analysis. It again showed the importance of perioperative critical care management to reduce mortality.
Age and Emergency surgeries also predicted both 90 days and in-hospital mortality independently. However, Age had weak strength of association with an odds ratio of 1.07 and 1.09 for 90 days and inhospital mortality respectively.
Logistic regression looked at 90 days mortality on multivariate analysis, we also analyzed time to the event by cox regression analysis which showed similar results, and age, emergency surgeries, and nontechnical complications were associated with worse survival over 90 days. Figure 1 showed on Kaplan Meier analysis non-technical complications were associated with signi cantly worse 90 days survival.
There are certain limitations of this study, as being a retrospective analysis there can be inherent limitations of retrospective analysis like selection bias. However, data was maintained prospectively. There can be other factors affecting mortality that could not be included in the analysis. The sample size was small as can be seen from the wider con dence interval. Our unit being predominantly Hepatopancreatic biliary surgery (HPB) unit in this analysis HPB surgeries outnumbered the other surgeries. The strength of the study is that it is one of the few studies showing strong strength of association with postoperative nontechnical complications with postoperative mortality.
In conclusion, Age, non-technical complications, open surgery, and emergency surgeries are independently associated with 90 days mortality, and age, acute kidney injury, non-procedural complications, and emergency surgery independently predicts in-hospital mortality.

Declarations
Ethical clearance: obtained from hospital ethics committee.
Competing interests: The authors declare no competing interests.  Figure 1