Chapter 3: Data Sources
3.4 Data from Other Sources
As discussed in Chapter 2, GPs face competition not only from other GPs, but also from hospital services, particularly emergency department and outpatient services. They also receive support from access to hospitals for more severely ill patients, and for advice and assistance from specialists.
In the Australian health system, outpatient departments are separate entities from emergency departments. The latter provide unreferred services to patients in an emergency, which may or may not lead to hospital admission depending on the nature of the medical condition. Outpatients departments on the other hand enable hospitals to provide a wide range of services to the community without patients being admitted to the hospital. These services include ante-natal clinics, specialist consultations, dialysis services, chemotherapy services and diabetes clinics. While some outpatient services may provide competition to GPs, no information is available activity levels of outpatient services.
3.4.1 Hospital Beds and Emergency Departments
Data were available on public hospitals (including bed numbers) and private hospitals (excluding bed numbers) from AIHW (Australian Institute of Health and Welfare 2000;
Australian Institute of Health and Welfare 2003).35 This source did not include information on the availability of emergency departments, and provided SLA information for public but not for private hospitals.
SLAs and bed numbers for private hospitals, and access to emergency departments, were drawn from the Hospital and Health Services Year Book which purported to
include all hospitals across Australia each year (see APN Business Information Group 2005, and other years).
Data on hospital bed numbers and availability of emergency departments were extracted for two years only (1998 and 2002), and linear interpolation and extrapolation used. Total hospital beds (public plus private, but excluding day surgery) and an indicator variable indicating the presence or absence of an emergency department in the SLA were derived. As the presence or absence of an emergency department changed in only six SLAs over the period, it was not practical to use this indicator in panel modelling, although it was used in cross-sectional modelling.
3.4.2 Nursing Home Beds
Nursing home patients were relevant to the GP market in that they were likely to require medical services (although overall only 1.5 per cent of GP services were provided in nursing homes in 2001).36 Large nursing homes at least were likely to change the measured level of mortality, particularly in small SLAs.
DoHA published lists of current residential aged care facilities electronically,37 but historical data were not available electronically. The Hospital and Health Services Year Book (e.g. APN Business Information Group 2005) purported to include a complete list
of nursing homes each year and, combined with DoHA lists for 2003, numbers of nursing home beds (approximately equivalent to high level care in the on-line lists) in
35 See http://www.aihw.gov.au/publications/hse/ahs02-03/ahs02-03-xd04.xls last accessed 16 October
2007.
36 Extracted from Medicare Australia on-line Medicare Statistics
http://www.medicareaustralia.gov.au/providers/health_statistics/statistical_reporting/medicare.shtml last accessed 16 October 2007.
1996, 2001 and 2003 were extracted and classified to SLAs from their postcodes. The numbers for intervening years were interpolated.
3.4.3 Health Status
There were two possible approaches to measuring health status, as data were available for mortality and for self assessed health. Section 3.6.1 shows that self-assessed health proved to be impractical for the purposes of panel analysis.
Mortality data were extracted at the SLA level for all years of the study (Australian Bureau of Statistics 2005). Crude death rates were preferred to standardised rates, as it is the total quantum of ill health which influences demand for health services.
The mortality data for each year were based on the SLA boundaries for that year, so concordances were used to convert the figures to 2001 SLAs. This translation was successful except for some 1996 data where clearly implausible results were derived, and the extreme results were replaced by imputed estimates.38
3.4.4 Concession Card Holders
Many GPs charge differentially for holders of concession cards and other patients, with the concession card providing a proxy for economic status. Data were obtained from the Department of Family and Community Services on the numbers of each of the concession card types (Pensioner Concession Cards, Health Care Cards, and
Commonwealth Seniors Health Cards) by postcode in September 2001 and September 2003. These data were converted to the SLA structure.
To obtain an SLA estimate of numbers of cards in other years, numbers were prorated within each category of cards according to the published totals (taken, for example from Commonwealth Department of Family and Community Services 2002), with allowance for population change by SLA. The total number of people covered by the cards in each
38 Extreme results for 1996 were identified as those which were more than 2 standard deviations from the
mean of the 8 observations for the SLA. They were imputed to be the mean for the SLA over the other 7 years, adjusted for the difference between the overall 7 year mean and the overall 1996 mean. All results were then adjusted to give the correct total deaths for 1996.
SLA was then estimated using national average number of persons per card for each card type (Commonwealth Department of Family and Community Services 2002). 3.4.5 Access to Private Schools
Richardson & Peacock (1999) suggested that the availability of private schools for their children may be a factor which attracted GPs to particular areas. Information on private schools funded by the Australian Government was available from the Department of Education Science and Training.39
These data provide a list of all funded schools, their suburb/town and level of funding. Testing with New South Wales schools showed correlations between numbers of GPs and total numbers of private schools in an area were negative. As the intuition behind using this variable was that GPs would be attracted to areas with wealthier schools, the list was restructured to include only those schools which were eligible for lower funding levels (defined as schools receiving less than 50 per cent AGSRC (Australian
Government Schools Recurrent Costs)). This redefinition, which excluded all Catholic systemic schools, led to a small positive correlation and was used in the analysis.