Profiles of institutional departments affect operative outcomes of eight gastroenterological procedures

Abstract Aim We evaluated the association of profiles of institutional departments with operative outcomes of eight major gastroenterological procedures. Methods We administered a 15‐item online survey to 2634 institutional departments in 2016 to investigate the association of questionnaire responses with operative mortality for the procedures. The proportions of conditions met were listed according to institutional volume and classified according to annual operative cases in 1464 departments. Group A included departments with annual performance of <40 cases of the eight procedures, B 40‐79 cases, C 80‐199 cases, D 200‐499 cases, and E ≥ 500 cases. We evaluated the number of conditions met for 10 of 15 items that could be improved by efforts of institutional departments, to assess whether the profiles of institutional departments had impacts on operative mortality. We built a multivariable logistic regression model for operative mortality with facilities categorized based on the number of conditions met and procedure‐specific predicted mortality as explanatory variables using generalized estimating equation to account for facility‐level clustering. We also examined how operative outcomes differed between facilities meeting nine or more conditions and those that did not. Results We recognized meeting nine out of the 10 conditions as being a good indicator for having appropriate structural and process measures for gastroenterological surgery. The facilities meeting nine or more of the conditions had better operative mortality for all eight procedures. Conclusions Our findings reveal that the profiles of institutional departments can reflect the outcomes of gastroenterological surgery in Japan.


| INTRODUC TI ON
In a previous study, 1 we reported that board-certified surgeons of gastroenterological surgery (BCS-Gs) contribute to favorable outcomes of gastroenterological surgery in Japan based on analysis of the National Clinical Database (NCD), a nationwide web-based data entry system for eight procedures of gastroenterological surgery, consisting of esophagectomy (Eso), distal gastrectomy (DG), total gastrectomy (TG), right hemicolectomy (RHC), low anterior resection (LAR), hepatectomy (Hx), pancreaticoduodenectomy (PD), and acute diffuse peritonitis surgery (ADP). In our previous study, the number of BCS-Gs in an institute was shown to have a significant correlation with operative mortality. To be specific, the ratio of observed to expected (O/E) operative mortality in institutions with four or more BCS-Gs was less than 1.0 for all procedures. Multivariable logistic regression showed that the number of institutional BCS-Gs was a predictor of operative mortality. As a result, we revealed that the number of institutional BCS-Gs is a surrogate marker of operative mortality.
In the present study, we investigated the association of profiles of institutional departments assessed by an online questionnaire survey with operative outcomes. The NCD, in which BCS-Gs are required to register their cases, commenced patient registration in January 2011. The gastroenterological surgery section of the NCD requires detailed input items for the eight major procedures. Using NCD data from 2011 regarding nationwide outcomes for the eight procedures, risk models of operative mortality [2][3][4][5][6][7][8][9] and morbidity [10][11][12][13][14][15][16][17] have been developed. These surgical risk models likely represent the current nationwide status in Japan because they are free of the patient selection bias that can occur in randomized controlled trials. The mortalities for all eight procedures seem acceptable as nationwide outcomes, being satisfactorily low compared with those reported in other countries. [18][19][20] Although requirements for application to become a boardcertified institute are authorized by the Japanese Society of Gastroenterological Surgery (JSGS), it has not been reported whether the profiles of institutional departments influence operative outcomes. In the present study, we investigated the impact of profiles of institutional departments on operative outcomes in Japan.

| Board certification system of the JSGS
The board certification system of the JSGS consists of boardcertified training institutions and BCS-Gs. Table 1 shows the requirements for JSGS-certified institutions. In Japan, there are approximately 2000 JSGS-certified institutions. Among 10 mandatory factors, having performed 120 essential major surgeries in the past 3 years is required of applicants for board certification.

| Data source and registry platform
The NCD was implemented in 2010 by 10 surgical societies including the Japan Surgical Society and the JSGS. Registration to the NCD through online data collection system was initiated in 2011.
This large nationwide database covers more than 95% of surgeries 1 Have performed 600 or more the gastroenterological surgeries determined by the Certified Committee (more than 120 of them essential major surgery a ) in the last three years.
2 Have a JSGS-certified two supervisory surgeons, or a BCS-G other than one supervisory surgeon.

3
Be capable of training for overall gastroenterological surgery. 4 Have a well-facilitated medical recording system in the institute.

5
Have an established ethical committee or be able to refer to other organizations when any ethics-related issues arise. 6 Have organized gastroenterology-related educational events (such as case conferences and mortality conferences) on a regular basis.

7
Have published more than three studies in any scientific journal or annual congress in the last three years. 8 Be capable of accepting physicians who wish to become a BCS-G. 9 Accept attendance at annual congresses or educational seminars as a part of training.

10
Be capable of rigorous investigation of the medical experience of applicants for the BCS-G.
Abbreviations: BCS-Gs, board-certified surgeons of gastroenterological surgery; JSGS, Japanese Society of Gastroenterological Surgery. a Surgery for esophageal cancer, distal gastrectomy, total gastrectomy, surgery for colon cancer, surgery for rectal cancer, surgery for bowel obstruction, partial hepatectomy, two or more segmentectomies of the liver, pancreaticoduodenectomy.

TA B L E 1
Requirements of application for a board-certified training institute authorized by the JSGS throughout Japan and more than 11 300 000 cumulative cases were registered by the end of 2018 21 . We used data from the NCD gastroenterological division for the present study.

| Questionnaire survey
We conducted an online questionnaire survey from February to March 2016 using the NCD system targeting all 2634 institutional departments that performed at least one gastroenterological surgery in 2015.
The 15 items in the questionnaire are shown in Table 2. The questions were created to capture the departments' structural as well as procedural characteristics on how they care for their surgical patients.

| Patients
We selected all patients who had undergone any of the eight proce-

| Classification of institutional departments based on annual case numbers
The strong associations between volume and operative outcomes have been reported elsewhere. [22][23][24][25] In the present study, the relationships between departmental annual case numbers of the eight procedures and departmental factors were investigated, to confirm the impact of the number of annual cases on operative outcomes. In accordance with the requirements for board certifi-

| Ten improvable departmental characteristics
Ten of the 15 items were selected to investigate the relationship between proportion of affirmative responses and operative outcomes. We chose these 10 items because they showed a positive association with operative outcomes and are conditions that can be improved by efforts of institutional departments or their institutes.

| Statistical analysis
For each question in the questionnaire, we assessed the number and percentage of departments by their responses. We also evaluated and reported using NCD gastroenterological data. [2][3][4][5][6][7][8][9]11 For cases that underwent more than two of the eight procedures, the risk model for the more invasive procedure was used for the estimation.
We assessed the responses to the questionnaire by the facility volume group as described above. To assess the association between number of conditions which the departments met among the 10 improvable items and operative mortality, we built a multivariable logistic regression model for operative mortality with facilities categorized based on the number of conditions met (0-1 being the lowest and 10 being the highest) and procedure-specific predicted mortality as explanatory variables using a generalized estimating equation to account for facility-level clustering. Furthermore, we also assessed the association of meeting nine or more of the conditions and operative outcomes by multivariable logistic regression analysis with each patient's baseline predicted risk as well as hospital case volume. This analysis was conducted among all patients, as well as by surgical procedures. We have presented the baseline variables used to estimate these predicted mortalities for each procedure, as in Appendix S1.
All tests were two-sided, and values of P < .05 were considered statistically significant. All analyses were performed using IBM SPSS version 24 (IBM Corp., Armonk, NY, USA).

| RE SULTS
Among 2634 institutional departments, 1579 responded to the questionnaire (59.9%). Among these, we selected 1464 institutional departments with at least one registered patient who underwent one of the eight selected procedures in 2015. As a result, 113 453 cases were included in the analysis.

| Differences in profiles of institutional departments across volume groups
The percentages of institutional departments with affirmative responses to the 15 questions across the category of annual case volume are shown in

| Association of institute volume category with operative outcomes
Mortality progressively decreased from Group A to Group E ( Figure 1). The mortality in Group E was 0.9%, and well below the average mortality of 2.5%. The same tendency was observed for the O/E ratio, in which Groups A and B had ratios of more than 1, and Groups C, D, and E had ratios of less than 1. The O/E ratio in group E was 0.49, which was remarkably low among the five groups. The adaptation of elective surgery is decided by preoperative conference.

| Improvable conditions and operative outcomes
Cancer Board is held.
MM conference is held.
NCD feedback system is used for clinical treatment.
Team treatment system is built.
ICT is installed.
NST is installed.
WHO safe check list is checked when starting a surgery.
There are two or more BCS-Gs.
There is certified nurse.
when compared with departments meeting all of the conditions. It also showed that while the departments meeting nine conditions had mortality odds that are not statistically significantly different from those meeting 10, the departments meeting only eight had statistically significantly higher odds ratio. Based on these findings, we recognized the use of meeting nine conditions among the 10 as being a good indicator for having appropriate structural and process measures for gastroenterological surgery. Figure 2 depicts the association between nine or more conditions met among the 10 improvable conditions and operative mortality. In the overall cohort, meeting nine or more conditions was significantly associated with decreased odds of operative mortality compared with facilities meeting eight or less conditions. When the association was assessed by procedure, the odds of operative mortality were significantly lower for all eight procedures.

| D ISCUSS I ON
The JSGS has a well-organized board system that has been maintained for a long time by board committee members. More than 6000 BCS-Gs have played central roles in gastroenterological surgery throughout Japan, but it has been difficult to evaluate the quality of BCS-Gs with regard to operative outcomes because of the small number of cases for which data were previously available.
Establishment of the NCD database has enabled us to accurately evaluate the contribution of BCS-Gs to better operative outcomes.
For the first time in our previous study, we demonstrated the positive impact of BCS-Gs on the outcomes of eight procedures of gastroenterological surgery. 1 In the context of our present study, we herein investigated the association of the profiles of institutional departments with

TA B L E 4 (Continued)
TA B L E 5 Relationship between annual case numbers of the eight major surgical procedures and affirmative response for each QI

Questionnaire item
Annual cases of the eight major surgical procedures F I G U R E 2 Impact of nine or more conditions met on mortality of eight procedures operative outcomes, using an online questionnaire survey comprising 15 questionnaire items (QIs) on the process for determination of surgical indications or surgical options, institutional department systems for patient safety, and improvement in quality of surgery.
We obtained responses from more than half of the institutional departments that received the questionnaire survey.
First, we investigated the correlation of affirmative response or negative response to every QI with operative outcomes, to identify individual items that were significantly positively related to operative outcomes. Thirteen items with affirmative responses among the 15 QIs had a significant positive impact on mortality.
We then focused on the association of institutional volume with operative outcomes, because some studies incorporating risk-adjusted models and using NCD data have investigated the association between institutional patient volume and operative outcomes for Eso, 22 DG, 23 Hx, 24 and PD. 25  Multivariable analysis showed that nine or more conditions met among 10 improvable characteristics was a predictor of operative mortality in all eight procedures. It is logical that high-volume institutes have a high number of conditions met, which was partly shown by the analysis of five categories classified by institutional annual cases. However, using multivariable analysis, we identified an effect of nine or more conditions met as an independent factor after adjustment for the effect of volume, which was significant.
Generally, institutional support systems, such as number of beds, being a teaching institution, and number of ICU beds, as well as various other profile elements established by the efforts by the institutions, including surgical conference, MM conference, cancer board, NST, ICT, and numbers of BCS-Gs and certified nurses, may contribute to better operative outcomes. However, "certified nurses" in this study included not only nurses for "perioperative nursing" or "critical care," but also those for "palliative care," "wound, ostomy and continence nursing," "cancer chemotherapy nursing," and others who might not directly influence mortality after surgery. Our throught is that the total number of several types of certified nurses is an important criterion to reflect the policy or culture for better medical care.
We want to place strong importance on clinical conferences, including preoperative, postoperative, and MM conferences, because they can be conducted at any department regardless of the hospital volume. Operative indications and operative methods should be discussed and determined by surgeons and physicians after precise evaluation of patient status, followed by open discussions of various aspects including curability, risks of complications, possible mortality, and age. Furthermore, an ICU and patient safety committee must be present and equipped in institutional departments where surgical treatment is commonly administered. In addition, the NCD feedback system can be leveraged to yield better patient prognosis.
In conclusion, this is the first report to demonstrate a positive association between profiles of institutional departments and operative outcomes. Institutional profiles play a large part in maintaining favorable outcomes of gastrointestinal surgery in Japan, as does the number of BCS-Gs.