Article Text

PDF
SP0136 Case Validation in Research Using Large Databases
  1. L. García Rodríguez1
  1. 1CEIFE, Madrid, Spain

Abstract

Computerized healthcare databases (CHCD) have been increasingly used in epidemiologic research and have become the single most used source of information in pharmacoepidemiology. A key feature in the selection of a computerized database for research is completeness and validity of recorded data. The Clinical Practice Research Datalink (CPRD - previously known as GPRD) and The Health Improvement Network (THIN) are two primary care UK databases that have an extensive track record of peer-reviewed studies. Yet, a large number of published research almost exclusively makes use of automated data. The validation process of a database is complex and the resources required to implement a study protocol using CHCD will vary widely depending on the need and amount of validation required. Validation process can be grouped under three sequential process: first, a computerized search to identify study population and patients with recorded outcomes: second, a manual review of patient profiles without and/or with free text information of primary care records: finally, a questionnaire sent to primary care physicians to confirm the validity of the studied outcome. Personel and time resources to complete these validation steps can be huge, but the quality and validity of the resutling research can be substantially improved. If researchers decide to skip this demanding validation process, then they should be ready to accept that journal reviewers and readers (when published) have legitimate doubts about the validity and relevance of their findings.

Disclosure of Interest None Declared

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.