Coding quality assurance

Clinical coding audit: what it is and who runs one in Australia

A clinical coding audit checks the quality of hospital coded data by having a second coder re-code a sample of records and compare the new codes with the codes originally assigned. Hospitals audit coded data because the codes drive activity based funding and national statistics.

Australia has no single national audit method. In the data quality statement for its Admitted patient care report (2026), the Australian Institute of Health and Welfare (AIHW) states that there are no national standards for coding auditing, so audit results cannot be used to measure coded data quality on a national basis. What follows draws on AIHW and state reporting, and on one published Queensland guide to audit practice. For the wider picture of coded hospital data, start with our guide to clinical coding in Australia.

A clinical coding audit is the independent re-coding of selected records and the comparison of the resulting codes with the codes originally assigned. That definition comes from the AIHW data quality statement accompanying the Admitted patient care 2024-25 report (2026). The Audit Office of NSW gives a second published description, in a 2015 performance audit: a different clinical coder reviews the code assigned to a patient’s admission and identifies whether an error has occurred through miscoding, and where miscoding is found, the coding auditor determines the severity of the error and its primary cause (Audit Office of NSW, 2015).

Coding audits are also a different exercise from clinical audits, which examine whether care met clinical standards rather than whether codes matched the record. Both matter in hospitals, but they check different things.

Three pressures keep hospital coding audit programmes going: funding, statistics and the Australian Coding Standards (ACS).

Funding comes first in the public system. The Independent Health and Aged Care Pricing Authority (IHACPA, formerly IHPA) states that Australian Refined Diagnosis Related Groups (AR-DRGs) are used to calculate public hospital funding on an activity basis, and that they are assigned largely from the codes clinical coders allocate, together with variables such as age and length of stay (2026). As the Queensland Coding Audit and Education Committee (QCAEC) guide notes, a DRG that changes after an audit can mean more or less funding for the hospital (2017). Our guides to AR-DRGs and clinical coding and hospital funding cover this in more depth. Under the 2017 Addendum to the National Health Reform Agreement, state governments were also required to give a statement of assurance on the accuracy and completeness of the activity data used for activity based funding reconciliation, a requirement the Australian National Audit Office described in its 2019 report on Commonwealth funding of public hospital services.

Statistics come second. IHACPA states that coding must be accurate and consistent with the guidance in the ACS to produce meaningful statistics (IHACPA, 2026). The AIHW publishes its Admitted patient care statistics from the National Hospital Morbidity Database, the collection its data quality statement covers (2026). Measuring coding accuracy against the Australian Coding Standards is among the purposes the QCAEC guide lists for hospital audit programmes (2017).

Responsibility for maintaining data quality during collection sits with data providers, the states, territories and hospitals, under IHACPA’s Data Quality Framework (2025). IHACPA (formerly IHPA) maintains the classifications and sets national prices. In practice, audit activity is organised at state and hospital level, and the mix of internal and external auditing varies.

In a quality assurance guide published in 2017 and hosted by the Clinical Coders’ Society of Australia (CCSA), the QCAEC draws a line between the two types. Internal audit programmes remain the domain of hospital based coding services and suit small, targeted audits that double as coder education, though targeted audits cannot provide a definitive measure of overall coding service quality. External audits suit larger samples that measure overall quality; because they are independent and structured, they reduce bias and allow benchmarking over time (QCAEC, 2017).

State and territory approaches

The AIHW appendix summarises how each jurisdiction organises its audit activity (2026):

Victoria’s Health Data Integrity Program has provided independent audits of unit record data since 1993, when the first audits mainly covered clinical coding of inpatient data under casemix funding (Victorian Department of Health, 2026, and its 2014 data integrity guidelines). Western Australia’s Health Information Audit Program measures DRG assignment, business rule compliance and coding accuracy; its patient activity compendium defines an audit as the independent examination of source data against reported patient activity (WA Department of Health, 2026).

The one published example framework, in the QCAEC guide (2017), applies nine criteria to each audited record: one for the principal diagnosis, and four each for diagnosis codes and procedure codes. This is one example from one Queensland guide, not a national standard.

• the wrong diagnosis was selected as the principal diagnosis under ACS 0001, which is about choosing the wrong condition and is assessed separately from picking the wrong code
• diagnosis codes that are incorrect, missing, overcoded or unsubstantiated, counted as four separate criteria
• procedure codes that are incorrect, missing, overcoded or unsubstantiated, another four criteria

“Overcoded” in that framework means the diagnosis did not meet the additional diagnosis criteria in ACS 0002, or the procedure was inherent in another coded procedure. “Unsubstantiated” means the code is not evident in the documentation. Procedure date, the condition onset flag and supplementary codes are also reviewed for accuracy in the same guide, reported separately from the nine criteria (QCAEC, 2017).

How errors are weighted and measured

Each error also gets a cause. QCAEC classifies causes as an ACS or coding rule failure, an ICD-10-AM convention or index and tabular failure, a documentation problem, or an abstraction error (2017). Errors are then weighted: a Type 1 error changed the DRG, a Type 2 error did not change the DRG but may show a failure to follow the classification, and a Type 3 error is minor. QCAEC is frank that this weighting is a subjective task at the discretion of the auditor, so it is not an objective national scale.

Two measures need to be kept apart. One is the code error rate, which QCAEC calls the error discrepancy rate: incorrect codes over the total auditor codes, multiplied by 100 (2017). Alongside it sits the DRG change rate, which counts audited episodes whose DRG changed. The two measures answer different questions, so a figure from one cannot stand in for the other.

Published findings are dated and local, and none amount to a current national error rate. Victoria reports that its state-wide AR-DRG change rate for audited records has been approximately 5%, a figure the state describes as indicating high-quality coded data, without stating the years covered (AIHW, 2026). South Australia’s 2011 coding audit found an AR-DRG change rate marginally above 10% for metropolitan hospitals, and the same appendix notes that a result under 10% is generally regarded as an indication of high-quality coding, a statement from South Australia rather than a national benchmark (AIHW, 2026).

What the older studies found

Two external audits run by the coding consultancy Pavilion Health covered 55 NSW and Queensland hospitals and 6,300 records in 2009 and 2010. The predicted DRG mismatch across the sample was 5.9%, and at one hospital with a mismatch of 5.6%, the case-weight change after correction still represented nearly A$8 million less in funding, caused by one error in a high-value DRG repeated many times (Reid and colleagues, 2011, a conference abstract by the consultancy that ran the audits).

In a blind re-coding audit of 752 surgical discharges at one Melbourne tertiary hospital over six months, approximately 16% of cases reflected a DRG change, equating to nearly A$575,300 in revenue, and 56% of the DRG change cases were due to documentation issues (Cheng and colleagues, 2009). Similar conclusions came from the Audit Office of NSW in 2015: common causes of miscoding include coder error and issues with clinical documentation, and coding accuracy relies on complete documentation. Because so many audit findings trace back to what clinicians wrote, our page on clinical documentation and coding accuracy follows that thread.

Treat these numbers as scoped to their own era, edition and sample, not as a rate you can expect at any particular hospital.

PICQ uses a set of indicators to assess the accuracy of clinical coding against the ACS, run over data that has already been coded (Victorian Department of Health, 2026). It was originally developed by the National Centre for Classification in Health (NCCH), and a 2008 study by Lam and colleagues judged it a useful initial screening tool after finding error types including completeness, redundancy, specificity and sequencing problems, with indicators graded as Fatal, Warning or Relative.

PICQ flags potential errors for a person to review; an audit re-codes sampled records from the full medical record, which the QCAEC guide names as the source document and the standard against which data quality is measured (2017). Edit and validation checks and PICQ screening are data quality activities; an audit means re-coding.

Where PICQ is used

Beamtree currently markets PICQ and states that it uses more than 900 indicators to assess coding accuracy, a vendor claim rather than an independent figure (Beamtree). Enterprise licences with Beamtree give Victorian public hospitals access to PICQ and RISQ, which assesses the accuracy of hospital-acquired complication coding (Victorian Department of Health, 2026), and the NSW Ministry of Health validates all public hospital coded data in that state with PICQ, benchmarking its Local Health Districts (AIHW, 2026).

A working coder takes part in a coding audit by having their own work audited and receiving feedback on the results. The QCAEC guide directs auditors to give constructive, evidence-based feedback to coders, treats feedback as essential because coded data will not improve until it is given, and has the audited codes corrected once the auditor and coder agree (2017). Conducting the audit itself sits higher: under the CCSA position statement on the role and duties of the clinical coder, conducting coding and related audits is among the higher-level duties normally undertaken by an advanced or senior clinical coder (2021). Advertised roles reflect that: a Gold Coast Hospital and Health Service auditor ad listed on SEEK in October 2026 asked for at least two years of hospital coding experience, and an iMedX ad that month asked for seven years of hospital auditing. “Clinical coding auditor” is a job title rather than a regulated or certified one.

Where coders first meet audits

Coders meet audits earlier, in training. In the HLT50321 Diploma of Clinical Coding, the final unit, HLTCCD007 Undertake moderately complex clinical coding, includes an element on participating in coding audits, with performance criteria covering preparing records for audit, validating your own coding and resolving error findings, and taking part in reviewing audit findings. Assessment requires at least one coding audit of the learner’s own work, and the knowledge evidence covers the purpose and types of coding audits, audit procedures and protocols, common coding errors and how to calculate error rates (training.gov.au, 2021). That unit trains a coder to take part in audits of their own work; it does not train someone to be a coding auditor.

What TalentMed offers

TalentMed (RTO 22151) does not offer a clinical coding audit course, and none of its five courses is one. BSB50920 Diploma of Quality Auditing is a general quality auditing qualification whose units contain no clinical coding content, and it should not be read as a coding audit pathway. For how coders move into these roles, see clinical coder career progression; for the published steps of an audit itself, see our page on the clinical coding audit process.