2.3.2. Data Analysis

Data were analysed in SAS Enterprise Guide 7.1 and Microsoft Excel©. Data visualizations were created in Tableau software version 2020.3. The data were presented as A3 colour posters, with data and graphics grouped by data source.

There are hundreds of mental health-related MBS items, and items are often added and removed from the schedule. The list of items analysed were similar to that described in Mental health services in Australia [21]. MBS data were analyzed by date of processing, as this results in more stable historical values over time. During code development, the extraction method was validated by reviewing results against those obtained through the Australian Government Services Australia Medicare Item Reports tool [22]. When extracting data, it was also ensured that data existed up to the last date of interest, to ensure that analyses were not impacted by unanticipated delays to data warehouse updates.

PBS items analyzed included antipsychotics (N05A), anxiolytics (N05B), hypnotics and sedatives (N05C), antidepressants (N06A), and psychostimulants, agents used for ADHD and nootropics (N06B), according to the Anatomical Therapeutic Chemical (ATC) Classification System [23]. PBS data were lagged by at least 6 weeks from the extraction date to reduce the effect of late claims, updates and cancellations.

Statistics supplied by non-governmen<sup>t</sup> organizations were routinely checked for consistency with historical supplies. Any changes to historical values were queried with the supplying organization, and then corrected if necessary. Time series were also routinely inspected manually for any anomalies or unusual patterns, which were also queried. Furthermore, these organizations may run their own data cleansing procedures from time to time, resulting in minor changes to historical values.
