Professor Gary Hulse
This randomised trial of adolescents presenting to hospital emergency departments in Perth with problems related to alcohol or other drug use, used the then recently implemented system of linking records in hospitals and medical practices in Western Australia. The aim of this research was to validate the use of hospital record linkage procedures as a method of collecting follow-up data and to use this methodology to assemble outcome data on the study cohort. As well as investigating the use of brief interventions with licit substance using adolescents and analyse hospital morbidity over 12 months, and the differences in the type of licit substances used by male and female adolescents, and also potential differences between single and poly substance users.
The project included a 5 year follow-up of participants who had enrolled in an earlier randomised controlled trial. The participants in that baseline study were adolescents who attended hospital emergency departments in Western Australia with conditions related to alcohol or other drug use.
In the baseline study, there was considerable attrition of participants. At 12 months post-intervention, only 69% of participants had agreed to be re-interviewed. It could be argued that those with the “worst” outcomes are less likely to agree to be interviewed, hence biasing the outcomes of the research.
However, by using the WA Linked Database, the authors were able to obtain 5 year follow-up data on 100% of participants. The work undertaken by this project was only possible because of the existence of the unique research infrastructure in Western Australia.
This study did not identify significant gender differences nor differences between users of alcohol alone and those who used alcohol in conjunction with other substances.
Data collection via the Emergency Department Information System provided a more complete data set than manual data collection. Manual extraction of information from hospital records is a labour intensive process which can be replaced by electronic extraction of information for most studies. Manual extraction may be justified when data are required on a small sample and particularly when the most update- information is essential.