- Research article
- Open Access
- Open Peer Review
Incomplete recording of Indigenous identification status under-estimates the prevalence of Indigenous population attending Australian general practices: a cross sectional study
BMC Health Services Researchvolume 19, Article number: 567 (2019)
Australian Aboriginal and Torres Strait Islander (Indigenous) peoples face major health disadvantage across many conditions. Recording of patients’ Indigenous status in general practice records supports equitable delivery of effective clinical services. National policy and accreditation standards mandate recording of Indigenous status in patient records, however for a large proportion of general practice patient records it remains incomplete. We assessed the completeness of Indigenous status in general practice patient records, and compared the patient self-reported Indigenous status to general practice medical records.
A cross sectional analysis of Indigenous status recorded at 95 Australian general practices, participating in the Australian Chlamydia Control Effectiveness Pilot (ACCEPt) in 2011. Demographic data were collected from medical records and patient surveys from 16 to 29 year old patients at general practices, and population composition from the 2011 Australian census. General practitioners (GPs) at the same practices were also surveyed. Completeness of Indigenous status in general practice patient records was measured with a 75% benchmark used in accreditation standards. Indigenous population composition from a patient self-reported survey was compared to Indigenous population composition in general practice records, and Australian census data.
Indigenous status was complete in 56% (median 60%, IQR 7–81%) of general practice records for 109,970 patients aged 16–29 years, and Indigenous status was complete for 92.5% of the 3355 patients aged 16–29 years who completed the survey at the same clinics. The median proportion per clinic of patients identified as Indigenous was 0.9%, lower than the 1.8% from the patient surveys and the 1.7% in clinic postcodes (ABS). Correlations between the proportion of Indigenous people self-reporting in the patient survey (5.2%) compared to status recorded in all patient records (2.1%) showed a fair association (r = 0.6468; p < 0.01). After excluding unknown /missing data, correlations weakened.
Incomplete Indigenous status records may under-estimate the true proportion of Indigenous people attending clinics but have higher association with self-reported status than estimates which exclude missing/unknown data. The reasons for incomplete Indigenous status recording in general practice should be explored so efforts to improve recording can be targeted and strengthened.
ACTRN12610000297022. Registered 13th April 2010.
Australian Aboriginal and Torres Strait Islander (hereafter Indigenous) peoples face major health disadvantage across a wide range of health conditions; including cancer, diabetes, and sexually transmissible infections (STIs) [1,2,3]. Reducing the persisting disparity of Indigenous peoples’ health is a key commitment of national health strategies and the Closing the Gap (CTG) framework [1, 4].
In Australia, more than half of the Indigenous population access healthcare through mainstream primary care services . Recording of a patients’ Indigenous status in practice records is important to ensure optimal health care and evaluate the effectiveness of clinical practices services and programs  recommended for Indigenous patients. For example, Indigenous adults are recommended to have annual health checks (Medicare Benefits Schedule (MBS) items 715 or 10,987) which provide preventive health strategies such as STI screening and detection of chronic disease risk factors ; and the CTG Pharmaceutical benefits scheme (PBS) co-payment reduces out-of-pocket costs of medicines for Indigenous people . Additionally, for health services, recognition of patients’ as Indigenous is required for health services to claim incentive payments, such as Medicare Australia’s Practice Incentive Program (PIP) Indigenous Health Incentives , which can support delivery of quality services. Finally, complete Indigenous status in patient files provides a more accurate picture for monitoring disease prevalence, to identify risk and plan effective services and programs .
National guidelines for the collection of Indigenous identification status in health data sets recommend the use of a standard verbal or written question (Fig. 1); and that service providers make completion of Indigenous status a mandatory requirement for new patient registrations . The Royal Australian College of General Practitioners (RACGP) [10, 11] recommends at least 75% of patient records be complete for cultural background where clinically relevant to achieve practice accreditation, including Indigenous status recording . A number of general practice audits have shown that a large proportion of patient records did not have an Indigenous status completed [12,13,14,15]. However to our knowledge no studies have compared the estimates of Indigenous population composition in general practice with other data sources, to provide an indication of how incomplete recording may under-estimate the true prevalence of the Indigenous population attending the clinics.
This study was completed in the context of the Australian Chlamydia Control Effectiveness Pilot (ACCEPt), a randomised controlled trial to determine if annual chlamydia testing for 16 to 29 year olds in general practice can reduce chlamydia prevalence. We took the opportunity to assess the completeness of Indigenous identification status recording (Aboriginal, Torres Strait Islander, both Aboriginal and Torres Strait Islander or neither) in general practice clinics, and examined factors associated with clinicians having greater than 75% completion in patient records. We also used a reference of patient self-reported Indigenous status to measure agreement of Indigenous status recording across administrative data sources, including complete and incomplete general practice medical records.
The ACCEPt trial was based in 52 towns in Victoria, New South Wales (NSW), Queensland (QLD) and South Australia (SA). All 152 primary care services (general practice and Aboriginal Community Controlled Health Services [ACCHSs]) located within these towns participated in ACCEPt; 134 were in regional and remote areas and nine in a major metropolitan city. The 2011 Australian Census reported that a higher proportion of Indigenous people reside in inner regional (22%) and outer regional /remote (30%) areas than in major cities (35%). NSW has the highest Indigenous population, around one-third (31%) of Indigenous people reside there, followed closely by QLD (28%) .
The study population used in these analyses was restricted to 44 of the 52 towns (95 clinics) with general practice clinics only, and which had data available for all data sources. We excluded all clinics located in towns (8 clinics) with an ACCHSs due to the known high completeness of Indigenous identification status recording in ACCHSs .
This cross-sectional study assessed completeness and agreement of Indigenous status recording in general practice using four data sources (described below); three from the ACCEPt trial and included i) general practice electronic patient records, ii) a patient survey, iii) a general practitioner survey  and the fourth was Australian Bureau of Statistics (ABS) population census data. Since data sources were independent and de-identified, patient self-reported Indigenous status was used to assess the agreement of Indigenous status recording compared to general practice and population data.
Patient medical records
Using strict confidentiality and privacy processes, de-identified information of all 16–29 year olds who attended 95 general practice clinics participating in ACCEPt during 2011 was extracted from patient information management systems using software called GRHANITE™ (‘GeneRic Health Network Information Technology for the Enterprise’) . Variables collated included routinely collected patient ID, clinic number, clinician identifier, gender, age, Indigenous identification status, and attendance date.
Between 2010 and 2012, a cross-sectional behavioural and biological patient survey was conducted to determine risk factors and the chlamydia prevalence in each town as they commenced ACCEPt. At each clinic, sexually active men and women aged 16–29 years were recruited consecutively to the survey when they attended for a GP consultation. The survey has been described in detail elsewhere . In brief, participants completed a self-reported questionnaire on a hand-held computer, including demographics (age, gender, Indigenous identification status), sexual behaviours, chlamydia knowledge, and clinic attendance information. Overall 4284 participants participated, with a response rate of 65.7% . These analyses use patient survey data from the 95 clinics described above (n = 3355). Completeness of responses to the Indigenous status question was 92.5%.
A cross sectional survey of GPs participating in ACCEPt was conducted during study recruitment meetings. GPs unable to complete the questionnaire at that time were given a reply-paid envelope to return the survey, with follow up reminders for non-responders. The survey details have been published elsewhere [20, 21]. In brief, self-administered, paper questionnaires captured clinicians’ demographics (gender, age), experience and training (country of primary medical degree and number of years worked in general practice), chlamydia knowledge, awareness and testing practices, including whether GPs routinely collected patients’ Indigenous identification status, and what proportion of patients the GPs thought were Indigenous. Of the 773 GPs enrolled in ACCEPt, there was a response rate of 72% overall . Responses from 496 (64%) GPs based at the 95 clinics described above were included in these analyses.
Australian Bureau of Statistics (ABS census)
ABS 2011 population census usual resident count data  were used to calculate the proportion of 16–29 year old Indigenous people for each of the 44 postcodes (95 clinics); and to determine the remoteness of clinics using the Australian Statistical Geographical standard (ASGS) classifications of major city, inner regional, outer regional, remote and very remote .
Completeness of Indigenous identification status
To assess the completeness of Indigenous identification status in the electronic patient records, the proportion of patients aged 16–29 years who had an Indigenous identification status recorded was calculated. We adopted a benchmark of 75% completeness of Indigenous status as the RACGP recommends that greater than 75% of patient records be complete for cultural background  and we stratified clinics as performing above or below.
To assess GP factors associated with high completion of Indigenous identification status (> 75% completion) in patient records, we used a univariate and multivariate logistic regression. Variables included in the model were remoteness, jurisdiction, GPs’ responses to routinely asking about patients’ Indigenous identification status; and GPs perceived proportion of Indigenous patients attending their clinic. Factors with a p < 0.05 in the univariate analysis were included in the multi-variate analysis.
Population composition of Indigenous people recorded in different data sources
We estimated the proportion of populations aged 16–29 years that identified as Indigenous for each data source: patient self-report, general practice patient records, and ABS postcodes. We calculated the overall proportion using the total number of Indigenous patients aged 16–29 years recorded across all clinics, as well as the median proportion and the interquartile range across all clinics. We focused on the median, due to the large variability in size of the clinics and the potential for larger clinics to influence the overall sample results.
For general practice records we used two different denominators; i) all patient records (includes those with a missing or unknown Indigenous identification status) and ii) patient records where those with missing or unknown Indigenous identification status are excluded (complete case analysis).
Agreement of Indigenous status recording across data sources
Since all data sources were de-identified it was not possible to use data linkage to assess accuracy of recording in general practice records. Patient self-report is the recommended approach for collecting Indigenous status in health data sets . Therefore the patient survey self-reported Indigenous status was used to measure the agreement of Indigenous status recording in the general practice and ABS data sources. We used a Pearson’s correlation to explain the direction and strength of the associations between self-reported Indigenous status and that found in the patient medical records (all patient records) and ABS census. We repeated the above correlations using only complete case patient records.
The study population included 95 general practice clinics located in NSW (28.4%), QLD (26.3%), Victoria (41.1%) and SA (4.2%). Clinics were located in major cities (9.7%), inner regional (72.6%), and outer regional or remote towns (17.7%). From the patient information management systems, we extracted records for 109,970 (clinic range: 60–12,120) patients aged 16–29 years old attending clinics in 2011.
Completeness of Indigenous identification status in patient records
Indigenous identification status was found to be complete for 61,086 (55.5% of all patient records aged 16–29 year olds (median of 60% per clinic) (Table 1). One-third of clinics (35%) had greater than 75% completion of Indigenous identification status in patient records.
In the clinician survey, 496 GPs from the 95 clinics provided a response; and 132 (27%) GPs were found to have greater than 75% complete recording of Indigenous status in their clinic patient records. GP characteristics and recording practices are listed in Table 2. A small proportion (5.6%) of GPs reported they always asked patients about their Indigenous identification status, 52.2% sometimes asked and 32.7% never asked. Most GPs (61.3%) thought that less than 1% of patients attending their clinic were Indigenous, with a further 17.9% reporting they were “unsure” of their clinics’ Indigenous attendees.
In the univariate logistic regression (Table 2), GP characteristics that were associated with > 75% completion of Indigenous identification status in 16–29 year old patient records were: GPs who reported always asking clients their Indigenous identification status’, and GPs’ perception that greater than 5%, or between 2 and 5% of clinic attendees were Indigenous. GP characteristics that were less likely to be associated with > 75% completion of Indigenous identification status were: clinics located in inner regional, and outer regional / remote areas; or those located in the States of Victoria or South Australia. These factors all remained significant in the multivariate model, except for GPs perceptions of the proportion of Indigenous clinic attendees (Table 2).
Population composition of Indigenous people recorded in different data sources
The population composition of Indigenous and non-Indigenous people for each data set is provided in Table 3.
Patient medical records
Overall, 2.1% of 16 to 29 year olds were recorded as Indigenous in patient medical records and the median proportion of 16–29 year olds recorded as Indigenous per clinic was 0.9%. The median proportion per clinic of non-Indigenous patients was 70.4 and 53.4% of combined clinic records. Using a complete case analysis of Indigenous status in patient medical records (44.5% n = 48,884), Indigenous people represented a higher median proportion per clinic of 3.4% among the 16–29 year old patient population, and 3.7% of combined clinic records.
In the patient survey, 3355 patients aged 16–29 years responded to the chlamydia prevalence survey, there was a 92.5% completeness of responses to the Indigenous status question. The clinic median proportion of patient identifying as Indigenous was 1.8%, and 173 (5.2%) patients overall.
The 2011 ABS census estimated a median proportion per postcode of Indigenous people aged 16–29 years in the 44 clinic postcodes as 1.7%, and overall was 1445 (1.6%) people.
Agreement of Indigenous status recording across data sources
Correlations between the proportion of Indigenous people self-reporting in the patient survey (5.2%) and all patient records (2.1%) showed a fair association (r = 0.6468; p < 0.01) (Table 4). After excluding records with an unknown and missing Indigenous status (complete case analysis), the correlation between weakened (r = 0.3136; p = 0.002) (Table 4).
The proportion of Indigenous people self-reporting in the patient survey was also correlated with ABS data (1.6%) showing a moderate association (r = 0.7010; p < 0.01) (Table 4).
This study found that almost half (44%) of patient records in the 95 general practice clinics had unknown or missing Indigenous identification status recorded. Using all patient records, the median proportion per clinic of patients identified as Indigenous was 0.9%, much lower than the 1.8% from the patient survey at the same clinics and the 1.7% in the clinic postcodes according to the ABS census data. The gaps in recording of Indigenous identification status are concerning but consistent with other studies [3, 14, 15]. However, to our knowledge this is the first study that assesses how incompleteness may bias the actual prevalence of Indigenous population attending general practice clinics through comparison with patient self-reported Indigenous status and population data.
We found that only 5.6% of GPs surveyed always asked patients about their Indigenous identification status, and these GPs were over four times more likely to have complete patient records. Additionally, GPs from regional and remote areas had poorer completeness, despite larger proportions of Indigenous populations residing in these areas . This finding differs to a previous study which showed regional and remote areas to have more complete records , and may have been influenced by better recording at the major city clinics included in our study. Health policy and GP training initiatives for Indigenous status recording in general practice should be strengthened in an effort to improve completeness, targeting the areas with known poorer recording. However, the success of such initiatives would require a thorough understanding of the barriers and facilitators to recording from the perspectives of both clinic staff and Indigenous people, including staff cultural awareness  and application in existing clinic systems [15, 23], appropriate ways to ask and explain why Indigenous status is being recorded  and willingness of patients to disclose due to fear of racial discrimination, social or historical factors .
In addition to delivering culturally safe care, Indigenous identification status is useful for disease monitoring and surveillance. Our study also showed that when complete case analysis was used for general practice records, there were poorer correlations with patient self-report than when all patient records were used; and the estimated median proportion of Indigenous people increased from 0.9% in all records to 3.4% in the complete case analysis. Therefore, population disease estimates relying on general practice data, which exclude the records with missing or unknown Indigenous status, are likely to bias and possibly over-estimate rates among Indigenous people. A Western Australian study using data linkage between an infectious disease surveillance database and established health administrative databases to improve accuracy of Indigenous identification status, found that rates of chlamydia and syphilis among Indigenous people had been overestimated by up to 30%, compared to the pre-linked rates which excluded cases of unknown Indigenous status .
There are a few potential limitations to consider in our analysis. First, the study was conducted in selected general practices and towns in four Australian States and may not be representative of clinics across Australia, particularly those in major cities. However, it remains important since Indigenous populations reside in these regional and remote areas  and across health care indices rates are higher for Indigenous populations in regional areas . Second, there is a reported undercount of Indigenous status for both Indigenous and non-Indigenous people in the 2011 census ; and further the ABS data focused on the entire postcode, and was compared to records for only patients who had attended the general practice that year, some of whom may have travelled from other postcodes. A study by Kong et al. showed that 74.8% of 16–29 year old Australians, attend a general practice at least once each year . However annual health care attendance would have only influenced our study if variations in attendance between Indigenous and non-Indigenous people exist.
In conclusion, around half the patient records in this study had missing or unknown Indigenous identification status, highlighting that further efforts are needed to improve completeness of Indigenous identification status in general practices. Complete and accurate recording of Indigenous identification status in patient records may allow health care providers to better address the healthcare needs of Indigenous people; and may enable better health policy decisions addressing the needs of Indigenous people.
Availability of data and materials
The datasets used and/or analysed during the study are available from the corresponding author on reasonable request.
Australian Bureau of Statistics
Australian Chlamydia Control Effectiveness Pilot
Aboriginal Community Controlled Health Services
Closing the Gap
GeneRic Health Network Information Technology for the Enterprise
Medicare Benefits Schedule
New South Wales
Pharmaceutical benefits scheme
Practice Incentive Program
Royal Australian College of General Practitioners
Sexually transmissible infections
Australian Health Ministers. Advisory Council. Aboriginal and Torres Strait Islander Health Performance Framework- 2012 report. Canberra: AHMAC; 2012.
Australian Bureau of Statistics. Australian Aboriginal and Torres Strait Islander Health Survey: First Results, Australia, 2012–13 2014. Report No. p. 4727.0.55.001.
The Kirby Institute. Bloodborne viral and sexually transmitted infections in Aboriginal and Torres Strait Islander people: Surveillance and Evaluation Report 2013. Sydney, NSW: The Kirby Institute, The University of New South Wales; 2013.
Australian Government. Closing the Gap- The Prime Minister's Report 2015. In: Department of the Prime Minister Cabinet- Indigenous Affairs, editor. Canberra: Australian Government; 2015.
Couzos S, Murray R. Aboriginal Primary Health Care: an evidence-based approach. South Melbourne, Victoria, Australia: Oxford University Press; 2008.
Australian Institute of Health and Welfare. National best practice guidelines for collecting Indigenous status in health data sets. Canberra: AIHW; 2010.
NACCHO/RACGP. National guide to a preventive health assessment for Aboriginal and Torres Strait Islander people. South Melbourne: The RACGP; 2012.
Hayman N. Improving Aboriginal and Torres Strait Islander people's access to the pharmaceutical benefits scheme. Aust Prescr. 2011;34:2). 841–4.
Medicare Australia. Indigenous Health Incentive 2014. Available from: https://www.humanservices.gov.au/organisations/health-professionals/services/medicare/practice-incentives-program. Accessed 10 Aug 2019.
Royal Australian College of General Practitioners (RACGP). Standards for general practices (4th edition. Melbourne, Australia; 2013 May 2013.
The RACGP National Faculty of Aboriginal and Torres Strait Islander Health. Identification of Aboriginal and Torres Strait Islander people in Australian general practice. 2011.
Britt HMG, Henderson J, Bayram C, et al. General practice activity in Australia 2012–13. Sydney, NSW. Sydney: University Press; 2013.
Thomson A, Morgan S, O'Mara P, et al. The recording of Aboriginal and Torres Strait Islander status in general practice clinical records: a cross-sectional study. Aust N Z J Public Health. 2015;40(Suppl 1):S70–4.
Ward J, Goller J, Ali H, Bowring A, et al. Chlamydia among Australian Aboriginal and/or Torres Strait Islander people attending sexual health services, general practices and Aboriginal community controlled health services. BMC Health Serv Res. 2014;14(1):285.
Schütze H, Pulver LJ, Harris M. What factors contribute to the continued low rates of Indigenous status identification in urban general practice?-A mixed-methods multiple site case study. BMC Health Serv Res. 2017;17(1):95.
Australian Bureau of Statistics. Estimates of Aboriginal and Torres Strait Islander Australians, June 2011. 3238.0.55.001. Canberra: Australian Bureau of Statistcs; 2013.
Hocking JS, et al. "12 PRT 09010: Australian Chlamydia Control Effectiveness Pilot (ACCEPt): a cluster randomised controlled trial of chlamydia testing in general practice (ACTRN1260000297022)." Lancet Accepted protocol summary. 2013. https://www.thelancet.com/protocol-reviews/12PRT-9010.
Boyle D, Kong F. A systematic mechanism for the ethical collection and interpretation of display format pathology test results from Australian Primary Care records. eJHI 2011: Vol 6(2):e:18
Yeung AH, Temple-Smith M, Fairley C, et al. Chlamydia prevalence in young attenders of rural and regional primary care services in Australia: A cross-sectional survey. Med J Aust. 2014;200(3):170–5.
Yeung A, Temple-Smith M, Spark S, et al. Improving chlamydia knowledge should lead to increased chlamydia testing among Australian general practitioners: a cross-sectional study of chlamydia testing uptake in general practice. BMC Infect Dis. 2014;14(1):584.
Lorch R, Hocking J, Temple-Smith M, et al. The chlamydia knowledge, awareness and testing practices of Australian general practitioners and practice nurses: survey findings from the Australian Chlamydia Control Effectiveness Pilot (ACCEPt). BMC Fam Prac. 2013;14(1):169.
Australian Bureau of Statistics. Census: Table Builder 2012. Available from: https://www.abs.gov.au/websitedbs/censushome.nsf/home/tablebuilder. Accessed Nov 2012.
Smith K, Fatima Y, Knight S. Are primary healthcare services culturally appropriate for Aboriginal people? Findings from a remote community. Aust J Prim Health. 2017;23:236–42.
Durey A, Thompson SC, Wood M. Time to bring down the twin towers in poor Aboriginal hospital care: addressing institutional racism and misunderstandings in communication. Intern Med J. 2012;42(1):17–22.
Mak DB, Watkins RE. Improving the accuracy of Aboriginal and non-Aboriginal disease notification rates using data linkage. BMC Health Serv Res. 2008;8(1):118.
Kong F, Guy RJ, Hocking JS, et al. Australian general practitioner chlamydia testing rates among young people. Med J Aust. 2011;194(5):249–52.
We thank: Jane Tomnay, from the Centre for Excellence in Rural Sexual Health, University of Melbourne for providing resources, support and advice on conducting this study in rural Victoria, and Dr. Douglas Boyle and the GHRANITE team, from the Health Informatics Unit, University of Melbourne; Carolyn Murray and Chris Bourne, from the Sexually Transmissible Infections Program Unit, NSW Health, for providing advice and resources. We acknowledge the support from pathology providers in providing chlamydia testing data for the analysis and thank Capital Pathology, Dorevitch Pathology, Douglass Hanly Moir Pathology, Healthscope Pathology, Institute of Clinical Pathology & Medical Research, Melbourne Pathology, Pathology North, Pathology Queensland, South Australia Pathology, St John of God Pathology, St. Vincent’s Health Pathology, Sullivan and Nicolaides Pathology, and Victorian Cytology Service. A big thank you to the ACCEPt research team of Alaina Vaisey, Michelle King, Eris Smyth, Jennifer Walker, Dyani Lewis, Lisa Edward, Chantal Maloney, and Anne Shaw for their help with recruiting clinics and providing them ongoing support.
We acknowledge input from the ACCEPt Consortium of investigators: Marcus Chen and Lena Sanci, from the University of Melbourne; David Wilson, and David Regan from the University of New South Wales; Sepehr Tabrizi, from the Royal Women’s Hospital; James Ward, from Baker IDI and University of New South Wales; Marian Pitts and Anne Mitchell, from La Trobe University; Rob Carter from Deakin University; Marion Saville, from the Victorian Cytology Service; Margaret Hellard, from the Burnet Institute; Nicola Low, from the University of Bern; and John Imrie, from the University of KwaZulu-Natal.
Consent to publication
This study was conducted as part of the Australian Government Department of Health - funded Australian Chlamydia Control Effectiveness Pilot. Additional funding and support has been provided by National Health and Medical Research Council, Victorian Department of Health and Human Services, NSW Health, Royal Australian College of General Practitioners, and the Australian Primary Health Care Nurses Association. The funding bodies were consulted about the design of the trial and were informed about the trial results but had no input to data collection, analysis and interpretation of the data, nor in writing the manuscript.
Ethics approval and consent to participate
The ACCEPt study was approved by the Royal Australian College of General Practitioners National Research and Evaluation Ethics Committee and the Aboriginal Health and Medical Research Council of NSW Human Research Ethics Committee. Written informed consent was obtained from all survey participants. General Practices provided administrative permission to install the GHRANITE software at their practice, and all General Practitioners enrolled in the study consented to the extraction of patient medical records recorded under their provider number.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.