Related Experiment Videos
Validation of the University of Manchester Drug Misuse Database
T Crabbe1, M Donmall, T Millar
1Centre for Urban and Community Research, Goldsmiths College, University of London.
Objective:
The study was conducted to assess the validity and quality of data held by one of the UK regional drug misuse databases (DMD).
Design:
The research was multi-centred and used retrospective analysis to assess the validity of data held on the database.
Setting:
The Regional Database is managed at the University of Manchester Drug Misuse Research Unit and uses data returned by medical and non-medical services within the UK's former North 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 drug teams (CDTs) during the course of 1993. Two datasets were used: the DMD dataset, based on returns to the regional database from the agencies in question; and agency client records. Additionally the data included on a random sample of 300 database forms returned by these CDTs were compared with information contained in client records.
Main Outcome Measures:
The study reports on how well DMD is functioning in relation to the correct reporting of episodes of problem drug use and the quality of data held.
Results:
A very high level of agreement (0.875 +/- 0.017, 95% CI, kappa coefficient 0.728) was established between reports sent in to the database and those expected by examination of agency records. The database figures underestimated the total number of episodes that should have been reported by a factor of 0.008. It was also established that 0.906 (+/- 0.018, 95% CI) of the reports made to the database were made correctly, that 0.178 (+/- 0.030, 95% CI) of eligible presentations were not reported, and that 0.166 (+/- 0.030, 95% CI) of ineligible presentations were mistakenly reported. Lastly, it was established that data were unnecessarily missing or inaccurately recorded 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 reliable, providing accurate measures of the extent and nature of presenting problem drug use in the region under study.