Download files
Abstract
Big Data has been used increasingly in pharmacoepidemiology, the study of medicines in large populations, where the majority of observational research uses population-based health data including pharmaceutical claims.
The Pharmaceutical Benefits Scheme (PBS) collects data on all medicines dispensed in Australia. This is an extremely important source of information for how medicines are used in the real-world; however, compared to other countries Australia has lagged behind in making best use of its data holdings for pharmacoepidemiology and pharmaceutical policy research, primarily due to the challenges of linking PBS claims with hospital, emergency department and other data collections.
This thesis demonstrates how to make best use of exposure data in the form of dispensing claims in the absence of outcomes data. We used a 10% sample of PBS dispensing data to evaluate four large-scale interventions and their impact on quality use of medicines in Australia, using robust and innovative statistical methods.
The four interventions include: (i) media coverage critical of statins; (ii) the re-scheduling of alprazolam to a “Controlled Drug”; (iii) the PBS-listing of cardiovascular fixed-dose combination products; and (iv) the introduction of a tamper-resistant formulation of controlled release (CR) oxycodone.
First, we found that the media coverage resulted in a decrease in statin dispensing, and an increase in statin discontinuation, even in people at high risk of cardiovascular events.
Second, we found that increasing restrictions on the prescribing and dispensing of alprazolam resulted in decreased prescribing and dispensing of alprazolam, and a decrease in alprazolam poisonings; but an increase in switching to other, less harmful, benzodiazepines.
Third, we found that fixed-dose combinations were associated with greater adherence in certain patient subgroups and poorer adherence in others, especially those not following prescribing guidelines, despite industry claims of benefits to all.
And last, we found that while the reformulation of oxycodone CR to be tamper-resistant led to a reduction in dispensing of high-strength oxycodone, there was also an increase in switching to other potent opioids in people most at risk of opioid misuse.
This thesis describes how standalone PBS data can be used to perform research that is timely, methodologically robust, policy-relevant, and internationally-relevant, even when outcome data are not available.