Development of an annually updated Japanese national clinical database for chest surgery in 2014

Objectives A national clinical database (NCD) adopted an “Internet-based collection” in 2011. An NCD specializing in chest surgery was launched based on the NCD system in 2014. The system was linked to the board certification as the second level in the hierarchy of the specialty of chest surgery and accreditation of educational institutions for chest surgery. Here, we report the status of the NCD for chest surgery in 2014 and clarified its registration rate and its accuracy. Methods Chest surgeries undertaken in Japan since January 1st, 2014 until the end of the same year were registered through an Internet-based system until April 8th, 2015. The registration rate was compared with the annual survey conducted by the Japanese Association for Thoracic Surgery (JATS) from 2011 to 2014. The rate of accurate inputting was measured by an Internet-based audit in reference to 563 anonymous operative notes of patients presented by 106 chest surgeons at the time of renewal for board certification for chest surgery. Results A total of 88,112 chest-surgical procedures were registered from 1000 chest surgery units (CSUs). Distribution of procedures by thoracic disease was almost identical to that of the annual survey conducted by JATS. However, the NCD had 4260 more registered procedures compared with the annual survey. The Internet-based audit showed that inter-rater agreement between Internet-based data and operative notes in any item was >94 %. Conclusions The NCD system can sustainably provide important and up-to-date information relating to preoperative status, oncology, and best practice for chest surgery in Japan.


Introduction
Clinical databases can be evaluated according to the quantity/quality of data, ''updatability'' and database size. Two surveys for chest surgery in Japan have been conducted: an annual survey by the Japanese Association for Thoracic Surgery (JATS) [1] and one by the Japanese Joint Committee of Lung-Cancer Registry [2]. Both were performed by mail from the flagship CSUs. The former is a conventional annual survey collected by JATS since 1986 and linked to the accreditation of educational institutions for chest surgery, and therefore, the survey has a highregistration rate. It focused only on the number of surgical procedures and their 30-day mortality and hospital mortality; few data on patient characteristics and perioperative evaluations are available. The annual survey by the Japanese Joint Committee of Lung-Cancer Registry has been conducted irregularly four times since 1994 and has focused on primary lung cancer. The size of that database accounted for fewer than half of the surgical procedures undertaken at the respective year in Japan. The quantity and quality of data on the oncologic characteristics of this database were of such a high standard that they were submitted to the staging committee of the International Association for the Study of Lung Cancer, and had a considerable impact on the seventh edition of guidelines for the staging of lung cancer. Each database is good in some aspects, yet poor in others. An ''ideal'' database has yet to be created.
An NCD in Japan adopted a ''web-based collection'' in 2011 with significant support from the Japan Surgical Society (JSS) [3]. Nine other surgical societies (Japanese Society for Gastroenterological Surgery; Japanese Society for Cardiovascular Surgery; Japanese Society for Vascular Surgery; JATS; Japanese Association for Chest Surgery (JACS); Japanese Society for Pediatric Surgeons; Japanese Breast Cancer Society; Japan Association for Endocrine Surgeons; and Japanese Society of Thyroid Surgery) joined the NCD. The NCD is a nationwide collaboration associated with the Japanese Surgical Board Certification System, in which data on 1.6 million surgical procedures from [4000 hospitals were collected in 2014.
An NCD specializing in chest surgery based on the existing NCD system was launched in 2014. The NCD for chest surgery required more detailed clinical data in addition to basic data. The NCD is linked to the second level in the hierarchy of specialty chest surgery and accreditation of educational institutions for chest surgery through webbased conversion, both of which are supported by the Japanese Board of General Thoracic Surgery.

Inclusion and exclusion criteria
Inclusion criteria of the NCD were surgeries via thoracotomy, thoracoscopy, or mediastinoscopy under general anesthesia. Exclusion criteria were: (1) surgery under local anesthesia; (2) bronchoscopic intervention with a tracheal stent; (3) percutaneous or endobronchial intervention in the thoracic cavity or mediastinum; and (4) removal of a stainless-steel metal bar after the Nuss procedure.

Data entry system
Chest-surgical procedures undertaken in Japan since January 1st, 2014 until the end of the same year could be registered until April 8th, 2015. Data were registered after patient data had been anonymized. Submission of patient data to the NCD (except for presumably, the very few (if any) patients who wished to opt out) was a prerequisite for the Japanese Surgical Board Certification System. Essential data were inputted at the first step for registration (Fig. 1). Surgeon-related data (identified by the license number of each surgeon registered by the JSS) were related to the surgeon's performance and referred to board certification at the first level of the hierarchy for surgery. The first step of registration could be completed if the operating surgeon did not switch to another field of surgery. The second step of registration started according to the field of surgery chosen as the second level of hierarchy. If ''chest surgery'' was selected as the second level of hierarchy, the detailed clinical data in relation to chest surgery were inputted until the completion of data registration without alerts showing inputting mistakes. Data were referred to board certification at the second level of surgery for chest surgery by the Japanese Board of General Thoracic Surgery and converted to annual survey data by JATS, if needed, which referred to the accreditation of educational institutes for chest surgery.

Data items for chest surgery
For the initial step in the registration process for chest surgery, the procedure is categorized as a ''surgical thoracic disease'', as a definitive diagnosis irrespective of the diagnosis upon hospital admission (Fig. 2). Disease category was identical to that for the annual survey by JATS, so that the NCD could be converted to the annual survey by JATS. All inputted items (which were displayed according to these thoracic diseases) had to be completed. Perioperative evaluations included ''common'' items and ''unique'' items according to the categorized disease (Tables 1, 2). Clinical and pathologic staging was inputted if the procedure was categorized as that for lung cancer. If procedurerelated complications occurred within 30 days after surgery, postoperative-complication items were registered even after hospital discharge (Table 3). If redo-surgery was necessary before discharge, the second procedure was registered as ''second surgery''. The patient was registered as ''second admission and the following surgery'' if the patient was discharged from hospital after the first surgical procedure and underwent redo-surgery. Finally, information relating to hospital discharge was inputted ( Table 4). The outcome was classified as ''in-hospital death'' even if the patient died after transfer to another hospital. The cause of surgical procedure-related death (including 30-day mortality and in-hospital mortality) was identical to that used for the annual survey by JATS and had to be inputted (Table 5).

Data registration
To assure the traceability of data through an Internet-based data-management system, directors of CSUs who were authorized by the NCD office approved of their data being registered in departments in charge of annual cases and responsible for data entry. They could revise web-based data whenever they noticed errors, especially at the time of conversion to the form of the annual survey by JATS, until April 8th, 2015.

Methods
The registration rate was compared with the annual survey between 2011 and 2014. Accurate input rate for certain items (age, sex, procedure, disease, operation time, blood loss, and surgeons) was measured by an Internet-based audit in reference to 563 anonymous operative notes of patients selected randomly by the NCD presented from 106 chest surgeons at the time of renewal for board certification for chest surgery. Authorized committees by the NCD and JACS could check inter-rater agreement between these operation notes and Internet-based data from the NCD office (Fig. 3).

Results
A total of 88,112 surgical procedures on the chest (including 38,999 procedures for primary lung cancer, 15,360 procedures for pneumothorax, and 4850 procedures for mediastinal tumor) were registered from 1000 CSUs to the NCD 2014 in the field of chest surgery. A total of 98.5 % of surgical procedures for a malignant pulmonary tumor, 95.5 % for a mediastinal tumor, and 93.3 % for bulla-related procedures were registered in the field of chest surgery, while quite a few surgical procedures for the pectus excavatum and palmer hyperhidrosis were registered in the other surgical fields (Fig. 4). Distribution of patient numbers according to each thoracic disease in the NCD for chest surgery was almost identical to the annual survey by JATS for the number of procedures categorized by thoracic disease (Fig. 5). The NCD had 4260 more registered procedures in comparison with the annual survey by JATS in 2014. More than 250 CSUs that had never provided data to the past annual survey registered data in the NCD. An Internet-based NCD in 2014 for chest surgery showed an inter-rater reliability of [94 % for each item (Fig. 6).

Discussion
An NCD for chest surgery was organized based on the NCD system to provide an ideal database for chest surgery.
A complete data-acquisition system linked with board certification for the second hierarchy of chest surgery and with accreditation of educational institutions for CSUs was successful and superior to the annual survey by JATS with  9 Interstitial pneumonia shadow on computed tomography of the chest 10 Child-Turcotte classification B or C respect to the number of registrations. The number of pulmonary resections for lung cancer registered in the NCD accounted for[95 % of all the cases registered in the Regional Bureau of Health and Welfare in Japan [4]. The system for NCD had a role as an eliminator of troublesome administrative procedures for chest surgeons and CSUs. This role may have been (at least in part), because JATS surveys are motivated by the intention to maintain institutional accreditation, whereas that of the NCD is by the individual qualification of application to the board of chest surgery. NCD can substitute for the annual survey by JATS in the future if data inputting can be more appropriate.
The registration system is dependent upon the field of surgery, because the NCD system is supported by JSS and nine other surgical societies. Procedures for pectus excavatum and palmar hyperhidrosis were not covered by the field of chest surgery. Pectus excavatum was mainly registered in the field of pediatric surgery for application to the board of pediatric surgery. The registration system should be unified regardless of fields of surgery in the future.
Lower accuracy tends to be a main problem for larger databases. Prevalence of inter-rater agreement through the audit system was satisfactory. In the future, Internet-based audit systems will be sustainable to maintain high quality in a large clinical database at reduced cost and reduced effort than site visits. The system will also be improved through greater experience and education on data input [5].
The goal of the NCD is to achieve high-quality surgical management [6]. To achieve this goal, the NCD will promote four sequential steps.
The first step is a risk model of pulmonary resection for lung cancer that is presented annually to each surgical unit for improving of surgical quality. Lung cancer is now a leading cause of death worldwide. Surgery remains the mainstay for complete cure, though other treatment modalities (e.g., molecular-targeted treatment or targeted  9 Required hemodialysis, postoperative serum concentration of creatinine [4 mg/dL or more than three times the preoperative value  radiation therapy) have been developed. Because of the aging society in Japan, many lung-cancer patients have comorbidities that can increase the risk of postoperative complications. The NCD can provide an annual estimated risk for preoperative status, surgical procedure, and oncologic status for each patient [7][8][9][10][11][12][13][14]. The NCD can be a Fig. 3 Internet-based audit system. An Internet-based audit system was developed to maintain the data quality of the NCD for chest surgery. Surgeons had to provide anonymous operative notes of patients selected randomly by the NCD and given to the JACS at the time of application for board certification for chest surgery. A committee authorized by the NCD and JACS could check the interrater reliability between these samples and Internet-based data from the NCD ''benchmark'' for lung-cancer surgery with several risk factors. The program of quality improvement for management of lung cancer can be sustainable and can provide feedback to CSUs, thereby eliminating ''cherry-picking'' problems associated with other programs [4]. Updating the NCD for chest surgery can ensure that the latest methods in the surgical treatment of lung cancer are disseminated. Follow-up surveys will be linked to the NCD system to show the long-term outcomes after lung-cancer surgery. A total of 38,999 procedures for primary lung-cancer patients undertaken in 2014 were registered, which were considerably higher than the numbers in the general thoracic databases reported by the Society of Thoracic Surgeons and European Society of Thoracic Surgeons [15,16] in terms of annual registration numbers. The impact on survival by surgical interventions will increase if the 5-year survival of these patients can be clarified by a follow-up survey. The second step is a risk model for other thoracic diseases. Further expansion of data as oncologic data in relation to relatively rare thoracic diseases (e.g., thymic carcinoma, mesothelioma) will be registered to clarify the effects of surgery on survival.
The third step will be a nationwide program for improvement in health-care and evidence-based policies. Data can show regional disparities in surgery for chest diseases. Identification of surgeons by the license number registered by the JSS can show the surgeon's performance for board certification but also their contribution to surgical outcomes [17]. Data on duration of hospitalization and operative costs (e.g., number of mechanical staplers) in relation to information on hospital discharge can aid costbenefit analyses. Such research will aid recommendations for nationwide improvement of health-care and evidencebased policies. The government should support a further research project using the NCD.
The final step will be risk profiles compared with overseas databases. The NCD will be compared with those from the Society of Thoracic Surgeons and European Society of Thoracic Surgeons if collaboration can result in standardization of nomenclature and variables among these databases. Such comparisons can lead to the improvement programs of surgical quality worldwide.
In the conclusion, an NCD system for chest surgery was well organized with a high-registration rate and high-accurate inputting rate. This was thanks to the elimination of troublesome administrative procedures for chest surgeons at the time of application for board certification and for CSUs at the time of application for institutional accreditation through presentation of the annual survey by JATS, respectively. The system can sustainably provide important and up-to-date information relating to preoperative status, oncology, and best practice in Japan. age sex procedure disease operation time blood loss surgeons Fig. 6 Ratio of inter-rater agreement of operative data based on 563 operative notes. Agreement was [94 % which is acceptable, but will improve with better education and database management