Development of computerized storage facilities for twin data: a relational database system for a twin register.

Abstract

Many twin registers hold information on flat file systems such as those provided by statistical packages or spreadsheets. Demographic details may be maintained separately from data collected in multiple different studies, leading to considerable problems with data consistency, redundancy, and integration. Ad hoc requests may be difficult. Implementation of a relational database system permits storage and maintenance of all records, simple data entry and validation procedures, linking of information from different projects with security of access, and the flexibility to provide rapid answers to ad hoc enquiries using standard Structured Query Language (SQL). Twin data provide a challenge for relational database design which rests on the technique of normalization and the use of unique identifiers to access associated groups of variables; for twins, "uniqueness" must preserve identification of both the pair and the individual twin subjects in the data structure to enable flexible access to and analysis of the data. An application on the Institute of Psychiatry Volunteer Twin Register (IOPVTR) database is described, through reference to one study of a sample of the twins, with simulated data. We show how a balance of adherence to database design principles and attention to ongoing clerical and research procedures has been used to produce an integrated, flexible, and open-ended system.