Using administrative data to look at changes in the level and distribution of out-of-pocket medical expenditure: An example using Medicare data from Australia

- Highlights gaps in the existing published statistics relating to out-of-pocket expenses.
- Provides a theoretical overview of the methods for developing out-of-pocket price index.
- Illustrates the utility of these statistics in disentangling trends out-of-pocket expenses.
- Discusses the potential for such statistics on developing and monitoring policy changes.

ObjectivesAustralia's universal health insurance system Medicare generates very large amounts of data on out-of-pocket expenditure (OOPE), but only highly aggregated statistics are routinely published. Our primary purpose is to develop indices from the Medicare administrative data to quantify changes in the level and distribution of OOPE on out-of-hospital medical services over time.MethodsData were obtained from the Australian Hypertension and Absolute Risk Study, which involved patients aged 55 years and over (n = 2653). Socio-economic and clinical information was collected and linked to Medicare records over a five-year period from March 2008. The Fisher price and quantity indices were used to evaluate year-to-year changes in OOPE. The relative concentration index was used to evaluate the distribution of OOPE across socio-economic strata.ResultsOur price index indicates that overall OOPE were not rising faster than inflation, but there was considerable variation across different types of services (e.g. OOPE on professional attendances rose by 20% over a five-year period, while all other items fell by around 14%). Concentration indices, adjusted for demographic factors and clinical need, indicate that OOPE tends to be higher among those on higher incomes.ConclusionsA major challenge in utilizing large administrative data sets is to develop reliable and easily interpretable statistics for policy makers. Price, quantity and concentration indices represent statistics that move us beyond the average.