Using Multiple Administrative Health Datasets to Identify A Mental Illness Cohort

Main Article Content

Margo Barr
Heidi Welberry
Julie Finch
Lou Anne Blunden


It is well established that mental illness is associated with poor physical health and chronic diseases possibly related to high levels of risk factors such as smoking, physical inactivity, poor diet and alcohol consumption. There is also evidence to suggest that having a mental illness may be related to poorer management of chronic diseases. Health service providers wanted to investigate this in Central and Eastern Sydney (CES).

Objectives and Approach
The purpose of this study was to scope the feasibility of identifying a mental illness cohort using CES residents from the 45 and Up Study (n=30,049) linked to administrative health datasets. These included Medicare Benefits Schedule (MBS) and Pharmaceutical Benefits Scheme (PBS) provided by Services Australia and hospital admissions provided by the NSW Centre for Health Record Linkage. We then compared differences in the identified cohort size, characteristics and the 8-year mortality rates.

Using only hospitalisation data, 6% of the CES cohort were identified as having a mental illness, compared to 17% using MBS data only, 26% PBS data only, 35%
using both the MBS and PBS data and 36% using all of the data sources. Crude mortality was 58% in those identified in the hospitalisation data, 27% based on the PBS data,10% using MBS data, 21% using the MBS and PBS data and 23% based on all combined sources.

Conclusion / Implications
We decided that the most appropriate option was to include the MBS, PBS and hospitalisation data to identify the mental illness cohort. This cohort will now be used to examine difference in the management of chronic disease, such as care plans and cycles of care, between those who do and do not have a mental illness.

Article Details

How to Cite
Barr, M., Welberry, H., Finch, J. and Blunden, L. A. (2020) “Using Multiple Administrative Health Datasets to Identify A Mental Illness Cohort”, International Journal of Population Data Science, 5(5). doi: 10.23889/ijpds.v5i5.1563.