|Page tools: Print Page Print All|
Assessing the feasibility of using data from the PBS
This section compares the self-reported medication use data from the 2014-15 NHS to the PBS subsidised prescriptions data represented in the MADIP asset to determine the suitability of replacing the medications questions module in future NHS cycles with the rich PBS administration data source.
These categories were chosen because the PBS supply date available in MADIP is represented only by month and year, meaning data comparisons could not be made specifically for the two weeks prior to the 2014-15 NHS interview. The 12 month prior period was of interest to consider for medicine use for long-term health conditions.
Of the linked 2014-15 NHS population, there was generally more people self-reporting medication use compared with people who had at least one script filled for a PBS subsidised prescription in the month prior to the NHS interview (46.7% compared with 41.6%). Greater similarity can be observed between the 2014-15 NHS and PBS subsidised prescriptions typically for ATC Classification System groups where medications are only available via prescription for example, Cardiovascular system (18.0% compared with 17.8% respectively). However, this pattern is not evident across all of the ATC Classification Systems groups and is particularly inconsistent where over-the-counter medications are typically used, for example, Blood and blood forming organs (7.5% compared with 3.9% respectively) see Table 1 below.
Table 1: Persons medication use self-reported in the 2014-15 NHS and PBS subsidised prescription(s) use
As expected, when the 2014-15 NHS two week self-reported medication use was compared to the PBS subsidised prescriptions filled up to 12 months prior to the 2014-15 NHS interview, a greater difference between the two data sources was observed for most ATC groups. However, where over-the-counter medications are readily used (for example, for Blood and blood forming organs, Respiratory system and Nervous System groups where aspirin, Ventolin, pain relief and cold medications are respectively used) the same pattern does not hold.
For information on over-the-counter and prescription medications in Australia, please refer to the Interpretation of results.
Agreement between 2014-15 NHS and the PBS subsidised prescriptions
To assess the impact of replacing the NHS medications questionnaire module with PBS subsidised prescriptions available via the MADIP asset, a proportion agreement between the two data sources was calculated as follows. The proportion agreement uses the frequency of agreements and disagreements observed for the linked records which have the presence (or non-presence) of a self-reported or PBS accessed medication in the month prior to interview, in a particular ATC group (i.e. a binary categorical variable).
Although test statistics such as Cohen’s ĸ could be used to statistically test the level of agreement between the two datasets, such statistics have several limitations, including difficulty of interpretation therefore the decision was made to report proportion agreement, which is directly interpretable.
The proportion agreement was calculated for each ATC Classification System group (see Table 2). High levels of agreement were achieved for most ATC Classification System groups when comparing self-reported medication use from the 2014-15 NHS to the PBS subsidised prescriptions represented in the MADIP asset.
Like the raw proportion comparisons in Table 1, levels of agreement were not as strong for ATC Classification System groups where over-the-counter medications are readily used, such as the Nervous system group.
Caution should be used when interpreting the level of agreement for the various ATC Classification System groups, as groups with a very small proportion of usage will inherently have a high proportion agreement. The proportion agreement should be viewed in conjunction with Table 1 to assess the overall comparability of the two data sources.
Overall, the linkage between the 2014-15 NHS and the PBS subsidised prescriptions as represented in the MADIP asset appears feasible to replace the survey data. For ATC Classification Groups where over-the-counter medications are commonly used, data could possibly be supplemented through specifically asking about the use of key over-the-counter medications in the survey.
Replacing survey content with the PBS subsidised prescriptions would help improve survey efficiency in future NHS collections. Whilst the PBS was not designed for research or planning purposes, the data are an objective and accurate record of medicines that were supplied and when they were supplied. Both data sources have pros and cons, but when taking into account the availability of existing data sources and respondent burden it appears feasible to use administrative data where possible as a survey data replacement. The ABS will endeavour to investigate whether additional administrative data sources exist that would help address over-the-counter medications, private prescriptions, dietary supplements and other data gaps.
After comparing time windows, it would be preferable to replace the NHS medications questionnaire module with PBS subsidised prescriptions as close to the survey reference period as possible. However, when analysing PBS subsidised prescriptions use, a longer time period would allow for more analysis of the patterns of medication use.
1. McHugh M. L. (2012). Interrater reliability: the kappa statistic. Biochemia medica, 22(3), 276–282 <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3900052/> accessed 21/04/2020.
2. Oksuzyan, A., Sauer, T., Gampe, H., Höhn, A., Wod, M., Christensen, K., and Wastesson, J. W. (2018). Is who you ask important? Concordance between survey and registry data on medication use among self- and proxy-respondents in the longitudinal study of aging Danish twins and the Danish 1905-cohort study. The Journals of Gerontology: Series A, Volume 74, Issue 5, May 2019, Pages 742–747 <https://doi.org/10.1093/gerona/gly104> accessed 06/03/2020.
3. Hafferty, J. D., Campbell, A, I., Navrady, L. B., Adams, M. J., MacIntyre, D., Lawrie, S. M., Nicodemus, K., Porteous, D. J., and McIntosh, A. M. (2018). Self-reported medication use validated through record linkage to national prescribing data. Journal of Clinical Epidemiology 94: 132-142 <https://www.sciencedirect.com/science/article/pii/S0895435617303062?via%3Dihub> accessed 24/02/2020.
These documents will be presented in a new window.
4364.0.55.015 - National Health Survey: Persons accessing Pharmaceutical Benefits Scheme subsidised prescriptions, 2014-15
Latest ISSUE Released at 11:30 AM (CANBERRA TIME) 09/10/2020 First Issue