Validation of the University of Manchester Drug Misuse Database

Objective-The study was conducted to assess the validity and quality of dat a held by one of the UK regional drug misuse databases (DMD). Design-The research was multi-centred and used retrospective analysis to as sess the validity of data held on the database. Setting-The Regional Database is managed at the University of Manchester Dr ug Misuse Research Unit and uses data returned by medical and non-medical s ervices within the UK's former Western Regional Health Authority. Material-The research was largely based on analysis of the reporting or non -reporting to DMD of 1526 presentations by drug users to four community dru g teams (CDTs) during the course of 1993. Two datasets were used: the DMD d ataset, based on returns to the regional database from the agencies in ques tion; and agency client records. Additionally the data included on a random sample of 300 database forms returned by these CDTs were compared with inf ormation contained in client records. Main outcome measures-the study reports on how well DMD is functioning in r elation to the correct reporting of episodes of problem drug use and the qu ality of data held. Results-A very high level of agreement (0.875 +/-0.017, 95% CI, kappa coeff icient 0.728) was established between reports sent in to the database and t hose expected by examination of agency records. The database figures undere stimated the total number of episodes that should have been reported by a f actor of 0.008. It was also established that 0.906 (+/-0.018, 95% CI) of th e reports made to the database were made correctly, that 0.178 (+/-0.030, 9 5% CI) of eligible presentations were not reported, and that 0.166 (+/-0.03 0, 95% CI) of ineligible presentations were mistakenly reported. Lastly, it was established that data were unnecessarily missing or inaccurately recor ded in 0.027 of cases and that data entry errors occurred in 0.015 of cases . Conclusions-The validation project showed that the DMD system is very relia ble, providing accurate measures of the extent and nature of presenting pro blem drug use in the region under study.