Following an evaluation of eight different methods of prevalence estimation the study team believe that the application of the following three approaches to Irish data will prove the most successful in estimating the prevalence of opiate use:

• Four sample Capture-recapture (CRC)
• Multiplier methods
• Multiple Indicator method (MIM)

Four sample CRC should be used to estimate prevalence of opiate use for all 26 counties. If this methodology fails to produce a county level estimate then either the MIM or multiplier methods should be used to calculate the estimate. These county estimates can then be summed to give regional estimates and an overall national estimate. It is recommended that the following four data sources should be constructed from the available data for inclusion in a four-sample capture recapture study:

• Central Treatment List
• HIPE
• Garda Síochána (PULSE)
• Probation Service data

We would also advise using the same matching criteria as previous Scottish and English studies. This would mean matching over initials, data of birth and gender. It is critical to the success of a new study to establish a geographical unit of analysis and work should be done to find the most suitable one prior to the beginning of any new prevalence estimation work. This unit of analysis needs to be exacting enough to help with the methodological difficulty of heterogeneity in highly populated areas but also mesh with current administrative geography to give estimates that are of practical use to policy makers and stakeholders. Following a roundtable discussion with prospective data providers, county of residence was identified as the preferred lowest level for estimates to be produced at. These county estimates can then be combined to give estimates at larger geographical areas such as Garda Síochána or HSE regions. This would mean taking into account counties such as Dublin and Tipperary which straddle these larger administrative areas.

In order to estimate the prevalence of cocaine use in Ireland, the study team recommend using a different methodology which combines large household survey data with longitudinal study data. This approach would account for the heterogeneity in the cocaine using population and would produce more than one estimate of cocaine use reflecting the different cocaine using cohorts among the population.