Estimating problem drug use prevalence at national level: Comparison of three methods

Abstract

Aims: Throughout the European Union efforts are made to obtain national estimates of problem drug use. These estimates are required for policy making. Given the importance of the estimates, much depends on their validity. Here we use two established methods for prevalence estimation, and compare it with a new one.

Method: Two established methods are the treatment multiplier and the multivariate indicator method. These are used to cross-validate the outcomes from a third method: multiple imputation. The latter has not been applied before in the context of prevalence estimation. Use is made of data from the Netherlands.

Results: The annual prevalence as estimated with the treatment multiplier is 0.30% (95%CI 0.24–0.36), and 0.31% (0.22–0.43) for the multivariate indicator method. The multiple imputation method came up with similar results: 0.32% (0.24–0.36).

Conclusion: All estimates compare well. The application of multiple imputation in this context is novel. It makes use of the same inputs that are needed for the multivariate indicator method, but has two advantages: it has a better way of handling statistical uncertainty, and the 95% confidence interval is readily available.

• Capture–recapture
• prevalence estimation
• opiate use