Logistic modelling of jurisdictional differences
States and territories take different approaches to treatment, both in terms of the mix of treatment types offered across treatment services, and how episode data from treatment episodes are recorded. One potential consequence of these differences is that clients in some jurisdictions may, on average, record more or fewer distinct treatment episodes. This in turn means that treatment may be more likely to be categorised as intensive in some jurisdictions than in others.
The AIHW has previously undertaken logistic regression modelling to collection period data between 2014–15 and 2019–20 to determine whether these jurisdictional differences affect national results by treatment cohort (AIHW 2021). To ensure accurate representation of the alcohol and other drugs treatment service client population, a similar logistic regression model was applied to AODTS collection period data between 2013–14 and 2020–21.
Modelling was undertaken to determine whether known treatment and reporting differences between states and territories were influencing the national results of clients receiving intensive, recurring and non-recurring treatment.
For example, while 23% of the clients in the AODTS NMDS recorded treatment in Victoria, 44% of clients receiving intensive treatment were based in Victoria (Table 6). However, it is important to note that Victorian data is not directly comparable with data for other jurisdictions as every treatment type provided is reported as a separate episode, regardless of whether it is a main or additional treatment type. This results in greater numbers of episodes compared to other states and territories, where main and additional treatment types are recorded under a single episode (AIHW 2022).
State/
territory
|
Intensive treatment (%)
|
Recurring treatment (%)
|
Non-recurring treatment (%)
|
All clients (%)
|
NSW
|
25.2
|
25.5
|
23.8
|
24.1
|
Vic
|
44.1
|
23.3
|
22.7
|
23.5
|
Qld
|
11.1
|
22.0
|
23.8
|
23.2
|
WA
|
8.9
|
17.7
|
15.6
|
15.7
|
SA
|
4.2
|
4.4
|
6.4
|
6.0
|
Tas
|
1.0
|
2.0
|
2.1
|
2.1
|
ACT
|
4.2
|
3.0
|
2.7
|
2.8
|
NT
|
1.5
|
2.1
|
2.8
|
2.7
|
Total
|
100
|
100
|
100
|
100
|
Differences between jurisdictions may affect the generalisability of the results. For example, if clients receiving AOD treatment in one state tend to be younger, and that state is more likely to have clients who received intensive treatment, then there may appear to be a relationship between age and intensive treatment that is purely caused by jurisdictional differences.
Modelling other client and treatment differences
To ensure differences between states and territories are not causing the differences between clients receiving intensive, recurring and non-recurring treatment in this report, a logistic model was applied. This model allows for exploration of the association between personal and treatment-level characteristics, while controlling for potential confounding effects between them. By controlling for state and territory in this model, it is possible to examine whether the other variables were still associated with intensive or recurring treatment.
The likelihood of a client receiving intensive or recurring treatment was modelled against the characteristics of the first two episodes that they received in the study period of 2013–14 and 2018–19 (noting that clients who first entered treatment from 2019–20 onwards are excluded, due to not meeting the minimum requirement of appearing in at least 3 collection periods).
After controlling for state and territory differences, the likelihood of a client receiving intensive or recurring treatment generally aligned with the results prior to controlling for state and territory differences.