**2. Background**

General practice forms a core part of the Australian healthcare system, often representing a patient's initial contact with the system. GPs in Australia manage patients across their lifespan, manage chronic health conditions and multimorbidity, and provide preventive healthcare. They also have a 'gatekeeper' role, providing referrals for patients to access other services including care from non-GP specialists, and subsidized care from allied health professionals for patients with chronic conditions. Currently, patients are free to

**Citation:** Gordon, J.; Britt, H.; Miller, G.C.; Henderson, J.; Scott, A.; Harrison, C. General Practice Statistics in Australia: Pushing a Round Peg into a Square Hole. *Int. J. Environ. Res. Public Health* **2022**, *19*, 1912. https://doi.org/10.3390/ ijerph19041912

Academic Editors: Richard Madden and Pentti Nieminen

Received: 15 December 2021 Accepted: 2 February 2022 Published: 9 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

attend one or more GPs of their choice, and are not assigned to a particular GP or practice. While patients have this freedom, most attend the same practice for continuity of care ([1], Chapter 15). In Australia, general practices are usually private medical practices providing "comprehensive, coordinated and continuing medical care drawing on biomedical, psychological, social and environmental understandings of health" [2].

In 2019, there were over 37,000 GPs in Australia, working across 8147 general practices [3]. According to data from the World Bank, 86% of the Australian population lived in urban areas in 2020 [4], primarily along the East Coast. Accordingly, in 2019, approximately three-quarters (74.5%) of full-time equivalent GPs reported working in major cities [3]. In any one year, approximately 87% of the population see a GP, and on average, there were six GP visits per head of population in Australia in 2015–2016 [1].

In March 2021, Australia had a population of 25.7 million people [5]. Funding of health services in Australia is the responsibility of the federal (national) and state/territory (regional) governments. Spending on health totaled \$197.7 billion (Australian) dollars in 2018–2019, equating to \$7772 per head of population [6]. Health spending represented 10% of gross domestic product.

In 2018–2019 \$65.5 billion was spent on primary health care [6], which incorporated general practice, allied and community health, and pharmacy (excluding Indigenous health care). General practice is primarily funded by the federal governmen<sup>t</sup> on a 'fee for service' model, where GPs can charge any fee they wish, and patients receive a fixed subsidy according to the Medicare Benefits Schedule (MBS), a catalogue of medical services for which a rebate can be claimed from the governmen<sup>t</sup> [7]. If the fee for a consultation or service provided is equal to the Medicare subsidy, then the consultation is 'bulk billed'. Around 87% of GP services are bulk-billed [3]. If not, then the patient pays an out-of-pocket cost decided by the GP. For patients with very high out-of-pocket costs for GP and non-GP specialist consultations, additional subsidies are provided through the Medicare Safety Net [8]. Medicare items for GP consultations are based on broad estimates of consultation length and complexity. Limited items are related to specific diseases or for specific population groups (e.g., annual health assessments for patients aged 75+ years, or chronic disease managemen<sup>t</sup> plans for patients with diabetes). Other Medicare-rebatable services include pathology tests, imaging tests and procedures undertaken. A separate Pharmaceutical Benefits Scheme (PBS) provides public subsidies for most prescribed medications dispensed by pharmacists [9].

The important role of general practice within the wider healthcare system has been recognized for some time. White et al. introduced a framework in Britain in 1961 to depict the organization of health care, demonstrating that within a population of 1000 adults, 250 (or 25%) will consult a physician (i.e., a primary care doctor or GP) in any one month. Nine of these 250 patients seeking care will be hospitalized, and five referred to another physician for care [10]. The overall stability of this framework has been established over time [11,12]. The aim of generating statistics from general practice is therefore not only to understand clinical activity undertaken in this setting, but to understand the health of the population overall.

In August 2019, the Australian Government released 'Australia's long-term national health plan'. The plan contained four 'pillars' (focus areas), the first of which was to strengthen the role of primary health care in the Australian healthcare system [13]. Later that year, a Primary Healh Reform Steering Group was established, focusing on the development of a ten-year plan for primary health care [14]. The draft report for the 'Primary Health Care 10 Year Plan' was released in October 2021 for consultation. The draft reforms are wide-ranging, containing changes to the funding models used in general practice, methods of general practice care delivery, and the introduction of patient registration at a single GP practice. The need for data to guide policy and quality improvement is reinforced in the plan [15].

If the draft reforms are implemented, there will likely be a multitude of changes to the current model of general practice in Australia. The proposed introduction of patient registration at a GP practice might further the role of the GP as central to population health. High-quality evidence-based statistics are required, to establish a baseline dataset for current general practice care delivery, and to assess the impact and effectiveness of any implemented reforms. This presents a timely opportunity to review the current state of general practice statistics in Australia.

#### **3. History of General Practice Statistics in Australia**

A detailed history of general practice data collection and analysis in Australia has been described elsewhere [16]. The first general practice survey was conducted by Dr Clifford Jungfer (GP) and Dr John Last (epidemiologist) in 1959–1960, with support from the (then) Australian College of General Practitioners [17]. This was followed by a National Morbidity Survey in 1962 [18]. Meanwhile, Dr Kevin Cullen, a GP in the town of Busselton, Western Australia, began the Busselton Health Study, a longitudinal study of population groups within Busselton conducted between 1966 and 1981. The Busselton Health Study was based on repeated cross-sectional surveys comprising questionnaires and blood tests to investigate the health of the study population, and identify health indicators that predicted future disease [19].

The Australian General Practice Morbidity and Prescribing Survey was conducted from 1969 to 1974, started by the Royal Australian College of General Practitioners' research committee, and led by Dr Charles Bridges-Webb [20]. The methods used in this study became the foundation for subsequent surveys of general practice clinical activity, including the Australian Morbidity and Treatment Survey (1990–1991) [21] and the Bettering the Evaluation and Care of Health (BEACH) study (1998–2016) [1].

For 18 years, the BEACH study described the clinical activity undertaken by GPs in Australia [1]. In BEACH, 1000 randomly selected GPs were sampled in each year of the study. Each GP participant recorded de-identified data for about 100 consecutive patient encounters on structured paper forms. Data collected included some patient demographics (e.g., date of birth, patient sex, postcode, Indigeneity), the patients' reasons for encounter (up to three), problems managed at the encounter (up to 4), medications prescribed/supplied/advised for purchase, for each problem, other treatments provided for each problem (including procedures and clinical treatments, such as advice and counselling), and pathology and imaging requests for each problem. Importantly, each managemen<sup>t</sup> action was explicitly linked to the problem for which that action was taken. More detailed methods for the BEACH study can be found elsewhere [1]. BEACH closed in 2016 due to the withdrawal of support from the federal governmen<sup>t</sup> (both funding and loss of the random samples of GPs provided) and wider losses of research support from industry partners [16]. With a final database spanning 18 years and approximately 1.8 million GP–patient encounter records, BEACH data were used to investigate the problems managed by GPs, how GPs managed these problems during consultations, and how the quality of care provided by GPs compared to evidence-based guidelines. BEACH data also identified changes in general practice clinical activity over time [22] and provided evidence about numerous policy areas, including time spent on patient care not able to be claimed through the MBS [23], the potential cost of freezing MBS item rebates [24] and (using length of consultation data) disproved statements that GPs were providing so-called 'six minute medicine' [25]. It was widely recognized that the closure of BEACH created a gap in data available about general practice [26]. Irving et al., in their investigation of primary care physician consultation time, presented a rather thorough international comparison of general practice data collection methods through their systematic review of 67 countries, and concluded that the Australian BEACH study "represents the gold standard for consistent reporting" [27].

The end of BEACH coincided with the closure of a number of other sources of data about general practice in Australia. Government funding was withdrawn from the Australian Primary Health Care Research Institute in 2015 [28]. The Medicine in Australia: Balancing Employment and Life (MABEL) study, a longitudinal study about the medical

workforce, ended in 2019, after 11 waves of data collection. This study provided numerous insights on access to medical care from between 3000 and 4000 GPs, followed up each year, including the drivers of hours worked, job satisfaction, and factors influencing recruitment and retention in rural areas [29]. The Australian Government's Medical Research Future Fund, established to provide grants for health and medical research, is reported to allocate less than 1% of total funding to primary care research [30]. Currently, many gaps exist in the statistics available from general practice, both in terms of the data collected and the research conducted [31].

#### **4. Current Status of General Practice Clinical Activity Data**

Limited administrative data are available about general practice from the MBS and the PBS. The MBS has records of the consultation items claimed by GPs from Government, but these provide very limited understanding of the clinical content of the consultation or the characteristics of the GPs. Similarly, the PBS contains data about subsidized medications dispensed by pharmacies, but does not include data about the clinical indication (i.e., symptom or diagnosis) for which the medication was prescribed. To obtain data about the clinical content of GP consultations, we need to look elsewhere.

General practice was one of the early adopters of computerized clinical records, with governmen<sup>t</sup> incentives to use computers available as early as 1998 [32]. Computerization began in the early 1990s, and some of the early systems developed (e.g., Medical Director) are still commonly used today [33]. There are now at least eight brands of electronic health records (EHRs) currently used in Australian general practice [33]. According to BEACH data, in 2014–2015, 97.5 % of Australian GPs reported that they used a computer for one or more purposes. However, only 70.7% used paperless medical records while 25.5% used hybrid (paper and electronic) records [22]. The MABEL survey in 2018 also asked about GPs' use of digital technology for a range of tasks, and found (for example) almost 90% of GPs using digital technology to view imaging pathology and results [34]. These data demonstrate that while GPs have a high uptake of computerized medical records and digital technologies, some still rely on paper for some activities.

While the BEACH study was conducted on paper, some GPs said they would have preferred to be able to download data from their practice electronic health records (EHRs) to be used in the study. There were two primary reasons that structured paper forms were used in BEACH. First, to facilitate the linkages between the problems managed and all managemen<sup>t</sup> actions provided for each problem. The problem–management linkage in BEACH ensured the GP specifically linked the prescription of a medication to the problem for which it was prescribed. It remains extremely difficult, if not impossible, to obtain these linkages from EHR data. This has led some researchers, using GP EHR data, to secondarily link each medication to a problem in the record on the basis of 'probability'. However, medications will often have multiple possible indications, let alone other off-label uses, making it difficult to know what health problem it is treating, and making matching by assumption highly unreliable. Second, BEACH was a study of GP clinical activity. The structured paper forms were inherently transportable, so that GPs who worked in multiple practices could take the forms between practices, or to home visits or nursing home visits. Secondary data entry by trained clinical coders, while time consuming and costly, facilitated consistent coding of the data to improve data quality.

In the absence of BEACH data since 2016, statistics from general practice have become focused on data extracted from EHRs. There are numerous research programs in Australia that rely on de-identified data extracted from GP EHRs, including:


Data extraction from EHRs may be as basic as a simple export tool. More complex extraction tools have been developed specifically for this purpose [35], for example GRHANITE (University of Melbourne) [36], the CAT4 tool (Pen Computing) [37] and

POLAR GP [38]. These tools can be used at multiple levels—for clinical audit or quality improvement activities at the practice, or by the local health region (called Primary Health Networks or PHNs in Australia), or to provide data to research programs at a wider level.

#### **5. The Use of EHR Data for Research and Statistics**

The automated extraction of data already collected during the clinical patient encounter creates a database of 'passive' data that can be used for statistics and research. While the primary purpose of data collection in an EHR is for patient care, making these data available for research and statistics minimizes the effort for individual GPs (who are often poor in time [39,40]) to participate in studies for multiple research groups. However, organizing and performing data extraction does involve time and effort for the practice. GPs report that it is often practice staff who undertake these activities [40], so the process is not entirely automated and does have a cost, although this is not always perceived as a barrier [41].

Passively collected data creates large volumes of data that can be interrogated in many ways. This provides greater scope to examine the managemen<sup>t</sup> of rare phenomena. Theoretically, for patients who regularly attend the same practice, EHR data extraction allows for the longitudinal analysis of a patient's journey over time, providing the potential to assess medical interventions and long-term health outcomes. This is limited though, if patients attend multiple practices (e.g., while travelling or for convenience) or change practices for any reason, resulting in incomplete data.

#### *5.1. Variability in EHR Design*

Interoperability of data requires standard approaches to data design structures, data field names and their associated definitions, and the coding and classification of relevant data fields. Standardization is required to enable data to be combined from different EHRs for clinical audits and research, and to facilitate the transfer of patient care between different healthcare providers (e.g., referrals). All of the GP EHRs used in Australia have been developed independently, which limits such interoperability and the ability to generate meaningful data from general practice EHRs, both for clinical and statistical purposes [33,35].

There are differences in the underlying designs of the EHR database structures, including the data field names, their definitions, and how data fields are or are not linked. There are also vast inconsistencies in the use of clinical classifications and terminologies, including the type of clinical terminology used (e.g., termsets developed by individual EHR developers, ICPC-2 PLUS [42] or SNOMED CT-AU [43]). In most EHRs, clinicians can choose whether to enter a term from one of these termsets or to enter free text [33]. As a result, most EHR research databases extract data from only some of the available EHRs, limiting the representativeness of the data. For example, MedicineInsight extracts data from the two most commonly used EHRs [44], each of which uses a different coding system.
