The development of the database can be classified into five stages.

1.

Planning

2.

Designing

3.

Implementation and operations

4.

Database governance policy, safety, and quality control

5.

Data analysis and utilization

Stage 1 – Planning

The field of oncology is vast and encompasses many varieties of cancers and various types of treatments which are multidisciplinary in nature. The lead clinician plays a crucial role in planning, development, and execution of clinical database program. He should have an in-depth understanding of the specialty and the anticipated utility of the database. IT team plays a supportive role in designing the software. The process of database development involves extensive brainstorming pertaining to database architecture, layout, format, content, input and output options, and additional utilities.

The information that needs to be captured will be different for individual cancers; e.g., breast cancer fields differ from the fields of head and neck or gastric cancer. Hence, it is recommended to develop database specific to individual cancers and the common fields for all cancers like demographics could be linked through relational databases.

Planning also involves related to software and hardware needs, networking, designing screen displays, access and security issues, and data analysis plans. There should be close collaboration between clinical and IT teams. Core clinical team should co-ordinate and onboard all clinical specialties including surgical oncology, medical oncology, radiation oncology, anesthesia, radiology, and pathology for a comprehensive and quality data collection. All the stakeholders should be clearly communicated regarding the data governance policy, joint ownership, authorship, and benefits of creating a comprehensive database.

Stage 2 – Designing

The next step is designing the general and disease-/organ-specific structured proformas with queries or fields including entry options. CCCDO layout can be divided into different workable modules including (1) demographics, (2) clinical presentation, (3) investigations, (4), treatment, and (5) follow-up. Output modules for generation of operation notes, discharge summary, audit reports, and census can be built-in to the program.

An ideal disease or organ-specific proforma should be 2 to 3 pages long and should contain 10 to 15 fields in each module. The format of fields or queries can be of 3 types—structured, semi-structured or unstructured, or open-ended. Structured queries are clear questions with well-defined options and the entry has to be selected from amongst the available dropdown options, e.g., gender—male or female. Similarly, e.g., to capture breast cancer tumor location, structured query created should be—laterality—left/right/bilateral, location – UOQ/UIQ/CQ/LIQ/LOQ. Unstructured queries are open-ended; i.e., the entry is not pre-conceived. Semi-structured fields have a defined option along with a text box to enter variable data. It is extremely important to plan and include as many structured fields as possible to standardize entry and facilitate coding for analysis. Treatment module should have separate sub-modules for surgery, radiotherapy, and systemic therapy. Surgical fields for each cancer will be unique; e.g., for extremity sarcomas, the fields displayed would be limb salvage surgery or amputation whereas for rectal cancer field displayed would be APR, sphincter salvage surgery (AR/LAR/ULAR/ISR). Standard and uniform database language should be used for designing queries, e.g., MeSH, ICD, and AJCC.

Every patient should be given a unique identification number. Front end of the database screen should display option for entering unique ID and tabs for various modules. Backend database will contain details of all data points. All the structured fields should be formatted on the screen with user-friendly drop-down menus. Tabs for generating discharge summary, operation notes, and census may also be provided.

The software should be user friendly, have a graphical user interface, should be robust, and have inbuilt safety features and options for data backup. The software should have an option to connect to web and be upgradable without loss of data. Database software should facilitate easy transfer of data to analytical and statistical programs such as MS Excel or SPSS. Softwares like Visual Basics, MS Access, MySQL, and Oracle can be used for database development [9]. The general layout of oncology database and the database modules are shown in Figs. 1 and 2.

Fig. 1

General layout of the clinical database at the Department of Surgical Oncology, AIIMS, New Delhi

Fig. 2

Showing format of database modules

Stage 3 – Implementation and Operations

A database can be established at different levels based on available expertise and resources. It can be developed and implemented at a hospital, department, unit, or consultant level. A comprehensive paperless database at the hospital level with networked computers is the option.

All stakeholders should be educated regarding the aims and objectives, standard operating procedures, data ownership, data sharing, data utilization, and authorship policies. Ideally clinical databases should be handled by treating doctors, e.g., residents/trainees rather than data entry operators under supervision of faculty. It should be integrated and implemented as a routine clinical activity. The basic minimum hardware required is a stand-alone PC with UPS and printer. For larger teams, 4 to 5 PC can be interconnected using a special purpose LAN network. The use of these PC for other purposes should be restricted. These PC should ideally not be placed on unsecured networks, to minimize chances of malware attacks and data loss. Use of USB should be restricted to authorized personnel only for data protection, safety, and privacy. The computer systems should be placed close to workplaces like outpatient clinics, wards, and operation theaters to facilitate easy end-user accessibility. The access should be restricted using passwords or biometrics. A simple three-stage entry at the time of workup and treatment planning, treatment delivery, and follow-up for each patient can be done in real-time basis.

Stage 4 – Database Governance Policy and Quality Control

For successful implementation of CCCDO, there should be a clear database governance policy covering following domains:

1.

Data entry protocols

2.

Patient privacy

3.

Data safety

4.

Data ownership

5.

Data sharing

6.

Data analysis

7.

Publication and authorship policies

8.

Quality control measures

All stakeholders should be informed and educated periodically. Some of the potential problems with the database include inconsistencies in data entry especially due to multiple users, errors in data entry, data corruption, and misuse of data.

Standard operating procedures (SOPs) should be defined and periodic audits by the database in charge should be conducted to check quality of entries, completeness of entries, and update follow-ups. During these exercises, we can generate census and perform mortality and morbidity and survival analysis. Regular backups of data should be taken as all computer systems are vulnerable and precious data may get lost due to software or hardware malfunctions.

Stage 5 – Data Analysis and Data Utilization

There should be a comprehensive plan for data analysis and utilization of data for academic and research purposes. There should be monthly audits for quality control and census generation. Annual data analysis can provide data regarding volume, spectrum of disease, and patterns of care. A morbidity and mortality analysis can also be performed annually to introspect and improve outcomes. A periodic survival analysis can be performed every 5 years.

As far as publication and presentation policies are concerned, there should be scope for every stakeholder to derive academic benefit from the database project. Residents can be made in charge of one specific organ-based database and they should be allowed to do dissertations related to that specific database. They should also be allowed to present the interim analysis in national and international conferences.

Major publications should be authored by faculty in charge of the organ system and should acknowledge the contribution of all contributors to the work and involve them as co-authors. All publications should be reviewed by all stakeholders to maintain transparency and honesty prior to submission.

AIIMS Surgical Oncology Experience of CCCDO

The CCCDO at the Department of Surgical Oncology at AIIMS, New Delhi, was initiated in 1997 by team of young faculty and residents. Extensive brainstorming and background work was done by clinical teams and a basic IT team was involved for software planning. General database modules and organ-specific modules were developed, and linkage established through relational database (MS Access). Database policies and governance model was developed and implemented. Databases were managed by trainees and senior residents and closely supervised by faculty. All trainees are given charge of one organ-specific database and also assigned a thesis topic related to that particular database. The CCCDO of AIIMS Surgical Oncology currently comprises of nearly two decades of clinical data of approximately 20,000 solid tumor patients. The data includes 5913 cases of breast cancer including 1000 BCS, 4500 head and neck cancers, 3325 gastrointestinal cancers (gastric, esophageal, and colorectal), 1645 bone and soft sarcomas, 1140 thoracic cancers, 1027 genitourinary, and 2488 mixed malignancies. Over years, the CCCDO has been an integral part of the academic program and research initiatives of the department. Due to the CCCDO, majority of the residents present papers based on original data and won numerous prizes and awards at national and international conferences. More than 250 scientific papers were published related to clinical and translational research [10].