Representativeness of participating GPs and patients
Each year the BEACH project tests the representativeness of their GP sample against all active GPs and their patient sample against all patients at GP items of service claimed through Medicare.1
Representativeness of the GP sample
The four years of BEACH data used in Healthy Communities: General practice care for patients with chronic conditions in 2009–2013 contain data from 3,903 participating GPs, representing approximately 19% of all practising GPs across Australia. To assess the representativeness of GPs in each Medicare Local catchment compared with all GPs in that catchment, the Authority obtained GP workforce data from the Department of Health and compared each sample of GPs in a Medicare Local catchment according to age, sex and place of qualification.
For the Medicare Local catchments with sufficient data to report (57 of 61), there were no significant differences according to sex for all 57 catchments, and reasonable representativeness according to place of qualification (49 of 57). There were differences in GP characteristics according to age for 27 of the 57 Medicare Local catchments largely due to the under-representation of GPs aged less than 35 years.
The Authority has determined that there must be at least 20 GPs represented in each Medicare Local catchment to publish results for that catchment. Six Medicare Local catchments had fewer than 20 GPs in the BEACH sample and were therefore not for publication.
Representativeness of the patient sample
To assess the representativeness of the patient sample, analysis was done on the size and age-sex disbribution of the sample.
The age-sex distribution of patients seen during GP consultations in the BEACH study for whom a Medicare rebate was to be claimed was seen to accurately represent the age-sex distribution of patients for whom GP service items are claimed through Medicare annually.1
The BEACH data reported includes data on 390,300 patient encounters. The sample size has a direct relationship with the standard error of the estimates provided, however this relationship is not linear. A minimum sample size of 250 patient encounters was determined by the Authority to ensure the precision of estimates, based on substantial improvements to the standard error as the sample size increases towards this number, then a flattening for larger sample sizes. A more conservative sample estimate was then chosen of 400 patient encounters and where there were between 250 and 399 patient encounters results were flagged as ‘interpret with caution’ to improve representativeness of the sample.
1. Britt H, Miller GC, et al. General practice activity in Australia 2012–13. General practice series no. 33. Sydney: Sydney University Press; 2013.