Development of a Value-Added Database of Evaluated Systematic Reviews in Veterinary Medicine: DVM Evidence
MetadataShow full item record
Objectives: This presentation will describe the development of the Database of Veterinary Medicine Evidence (DVM Evidence). The database was created by collecting, annotating, and evaluating systematic reviews and meta-analyses of relevance to veterinary medicine. Methods: The process includes four steps, similar to conducting a systematic review: identification, selection, appraisal, and data abstraction. Identification includes searching bibliographic databases (MEDLINE, Cab Abstracts, and more), grey literature, and expert selected list of conference proceedings and journal titles. The selection and appraisal process was completed independently by 2 team members, with disagreements settled by consensus. The appraisal process utilized AMSTAR and PRISMA. The data abstraction form includes citation, review question, inclusion criteria, topic, resources searched, and list of included primary studies. The database will be open access and browsable by species, specialty area, and type of study as well as searchable by keywords, author(s), journal titles, and year. Collaboration will be sought with librarians, researchers, and veterinarians to add in various perspectives. Results: A pilot group of 20 studies was selected by the team’s systematic review expert based on criteria designed to facilitate training team members in appraisal, coding, and testing the coding form. The coding form which incorporates AMSTAR, PRISMA, and custom questions, was developed in Qualtrics, a subscription based survey tool. Although initially the survey format of Qualtrics had some promise as a tool for the coding form, several issues were found during the pilot phase. The team decided to build a custom form that provides more flexibility with importing and exporting data to and from the database, and more control in the desired outcome of the resource. The team chose MySQL for the database management system, and PHP as the scripting language, because these software are supported by the library IT department, powerful and flexible enough to accomplish the tasks, and common enough for others to adapt them as the database evolves and personnel change occurs. Graduate students, selected for their expertise with MySQL and PHP, created the database and public interfaces with design input from team members. During each step of the process, the team considered needs of potential external evaluators. Conclusions: Developing a systematic review database of this complexity takes a team with a variety of skills, strategic planning, and frequent checkpoints along the way. The next step will focus on the usability of the custom made form and the search interface with external evaluators.