Defining dementia and fall cohort
Dementia identification
People were identified as having dementia in this study where they had a record of one or more of the following:
- dispensed a dementia-specific medication through the Pharmaceutical Benefits Scheme (PBS) between 1 July 2010 and 30 June 2020
- hospital admission between 1 July 2010 and 30 June 2020 with a principal, additional or supplementary code diagnosis of dementia
- emergency department presentation between 1 July 2010 and 30 June 2020 with a principal or additional diagnosis of dementia
- Aged Care Funding Instrument (ACFI) assessment between 1 July 2010 and 30 June 2020 with a mental or behavioural diagnosis of dementia.
Using linked information from multiple datasets over a number of years means that anyone whose dementia diagnosis was not recorded in one dataset at any point in time (for example, during a hospitalisation), may be recorded in other datasets and/or at another point in time.
The dementia cohort for this study comprises 132,150 people aged 30 and over who had a dementia-related record in the National Health Data Hub (NHDH) and met the criteria outlined below in ‘Inclusion criteria for study cohort’. This cohort is a subset of people living with dementia and does not account for all people living with dementia in Australia.
The exact number of people living with dementia in Australia (the ‘prevalence’) is currently not known and estimates vary widely (Dobson et al. 2023). The Australian Institute of Health and Welfare (AIHW) dementia prevalence estimates are derived from a variety of data sources of varying quality, including small-scale Australian and international epidemiological studies. It is not currently known how dementia prevalence rates vary between geographic areas in Australia.
The dementia study cohort only includes people with a record relating to dementia in one of the administrative data sets in the NHDH (such as data on medications dispensed, public hospital admissions, residential aged care assessments and causes of death).
Hospital data are an important source of dementia information, so to ensure comparability of findings across regions, people from states/territories without hospital data in the linked data (Western Australia and the Northern Territory) were excluded from the cohort.
Most administrative data sources for dementia relate to later stages of disease, such as hospital stays, residential aged care and death. Older people with comorbidities are therefore more likely to be identified in the study cohort, while younger people and people in the early to moderate stages of dementia are less likely to be identified. People living in rural and remote areas are also less likely to be identified in the linked data, as they often access specialised aged care services (AIHW 2023a) for which data are not available in the NHDH. It should also be noted that dementia is often underreported on death records (AIHW 2023b; Xu et al. 2022).
Severity of dementia cannot be ascertained from the available administrative data, although it is likely that identification of dementia from hospital and aged care data largely captures moderate-severe dementia.
Whilst types of dementia (such as Alzheimer’s disease or Fronto-temporal dementia) can be recorded in hospital data, they are not recorded in detail in aged care data and as such, individuals who do not have dementia-related hospitalisations will not have data regarding the specific type of dementia they have. Further, the type of dementia identified may not be recorded correctly, may change over time or an individual may be diagnosed with multiple types of dementia. Due to these limitations, type of dementia was not analysed in the present study.
Defining the fall cohort
To examine the risk factors associated with first hospitalised falls, and the characteristics and outcomes of those falls, the study centred on people who had their first hospitalised fall in 2019, with a comparison group comprised of individuals who did not have a hospitalised fall prior to or during 2019. Individuals without a hospitalised fall were allocated a randomly generated reference date between 1 January 2019 through to 31 December 2019, or their date of death, whichever occurred first, to be used when comparing outcomes. This is referred to as the ‘no fall reference date’.
While falls that do not result in hospitalisation occur often among people living with dementia, there is no standard method for these to be recorded, particularly within the community population. Data on falls in residential aged care are recorded in the Residential Aged Care Quality Indicators – January to March 2024. For example, in January to March 2024, 31.3% of care recipients had a fall recorded, and 1.8% had a fall that resulted in major injury (AIHW 2024). However, these data are not available at the individual level in the NHDH. This study therefore focused on falls that resulted in a hospitalisation as this allowed person-level outcomes and risk factors to be analysed.
The year 2019 was chosen because it was the latest year in the data which still allowed 12-month outcomes to be examined.
People were included in the study if they:
- had a dementia record in the data that preceded or occurred within a year following the fall (or no fall reference date) up to 30 June 2020
- had their first hospitalised fall in 2019 (fall cohort) or had no prior record of a fall up to 31 December 2019 (no fall comparison cohort). Note that the dataset only contains records from 2010 onwards so any hospitalised falls prior to this could not be captured
- lived in NSW, Victoria, Queensland, South Australia, Tasmania, or the ACT in the period 1 January 2019 to 31 December 2020. The study was limited to jurisdictions with complete public hospitals data available in the linked data. Individuals whose last place of residence prior to 2019 was in Northern Territory or Western Australia were excluded from analysis
- had complete information on key variables of interest in the linked data, such as age and sex
- were aged 30 years or older in 2019. This age cut-off includes most people with younger onset dementia whilst ensuring adequate numbers for analysis. Childhood dementia was out of scope for this study
- had an MBS record within the 12 months preceding the reference date (fall or no fall assigned date)
- had a PBS record in the available data.
Place of residence was determined for each person in the cohort, based on the available administrative data.
- People living in residential aged care includes people who had an episode of permanent or respite residential aged care that surrounded or ended on the admission or no fall reference date.
- People living in the community includes people without an episode of residential aged care that surrounded or ended on the date of admission or no fall reference date. This may include people who were living in other supported accommodation or residential aged care facilities not included in the linked data.
AIHW (2023a) Providers, services and places in aged care, AIHW, Australian Government, accessed 26 July 2024.
AIHW (2023b) Dementia in Australia, AIHW, Australian Government, accessed 28 June 2024.
AIHW (2024) Residential Aged Care Quality Indicators — January to March 2024, AIHW, Australian Government, accessed 16 September 2024.
Dobson AJ, Flicker L, Almeida OP, Waller M, Anstey K (2023) ‘Different estimates of the prevalence of dementia in Australia, 2021’, Medical Journal of Australia, 218, 320-321, doi:10.5694/mja2.51838.
Xu Z, Hockey R, McElwee P, Waller M, and Dobson A (2022) ‘Accuracy of death certifications of diabetes, dementia and cancer in Australia: a population-based cohort study’, BMC Public Health, 22(1), 902, doi:10.1186/s12889-022-13304-8.