Documentation and coding accuracy

Clinical documentation improvement in Australia: why documentation drives coding accuracy

Clinical documentation improvement in Australia is the work of making clinical records complete and specific enough for a clinical coder to code every admission accurately, and it matters because documentation is the only thing a coder codes from. Every diagnosis and procedure code on an admission must be supported by the record. When documentation is missing, vague or conflicting, a condition may be coded at a lower specificity, missed entirely, or coded without evidence behind it, and those gaps surface later as coding errors and audit findings. In its 2015 performance audit of activity based funding data quality, the Audit Office of NSW found that common causes of miscoding include coder error and issues with clinical documentation, and that coding accuracy relies on complete documentation.

This page covers where documentation causes coding errors, how documentation queries work, why hospital-acquired complications depend on coded data, and what clinical documentation improvement looks like in Australian practice. If you are new to the discipline, start with our guide to clinical coding in Australia.

Documentation gaps surface in coded data as unsubstantiated codes, overcoded additional diagnoses and wrong principal diagnosis selections, and auditors record documentation itself as a cause of coding errors. One published Queensland guide, the Queensland Coding Audit and Education Committee’s Suggestions for Quality Assurance (2017), maps exactly that ground (QCAEC, 2017). Its example audit criteria include diagnosis or procedure codes recorded as unsubstantiated, meaning the code is not evident in the documentation, and additional diagnoses recorded as overcoded because they did not meet the ACS 0002 criteria. Its principal diagnosis criterion is the wrong condition selected under ACS 0001. It classifies the cause of each error as a failure against the Australian Coding Standards, a failure against ICD-10-AM conventions or the index, an abstraction error, or documentation that is “inadequate, poorly constructed, conflicting, missing or illegible”.

Principal diagnosis selection is a documentation-sensitive step of its own, because a coder can only select the principal diagnosis from what the record documents. For that selection in detail, see our guide to principal diagnosis in Australia, and for the ACS 0002 criteria see the guide to coding comorbidities and additional diagnoses.

The table below collects the study evidence for this page. Every study in it describes a local setting, whether one hospital, two hospitals or one health service, so none of the results is a national error rate.

A documentation query is a structured question from a coder or a coding auditor to the treating clinician, asking the clinician to clarify, confirm or complete something in the record so the episode can be coded correctly. Rather than guessing at an ambiguous note, the coder asks.

That same Queensland guide (QCAEC 2017) says differences in clinical interpretation between a coder and an auditor are best resolved with a documentation query to the treating clinician (QCAEC, 2017). Audit counting handles queries carefully too. Where an auditor notes that a documentation query might have assisted code selection, the query is discussed with the coder and reported separately; it is not counted as a coding error. Where a code is actually wrong, by contrast, documentation can be recorded as the cause of the error.

The guide lists documentation issues as a standard item in audit reports to hospital executives, and its documentation causes include cases where a documentation query would have helped the coder allocate a better code (QCAEC, 2017).

Hospital-acquired complications (HACs) are where documentation gaps reach beyond the coding department: the Australian Commission on Safety and Quality in Health Care (ACSQHC) identifies them from ICD-10-AM codes combined with the condition onset flag, so HAC reporting depends on accurate coding and documentation (ACSQHC, 2025). Neither input exists independently of documentation: the code reflects what the record says, and the condition onset flag is assigned from what the record supports. Because hospital-acquired complications are identified from the condition onset flag, one published Queensland guide reviews the flag for accuracy in its example audit process, reported separately from its nine error criteria (QCAEC, 2017).

The same coded data feeds hospital funding. The Independent Health and Aged Care Pricing Authority (IHACPA, formerly IHPA) introduced a funding adjustment for episodes with a HAC in July 2018, according to its 2026 fact sheet (IHACPA, 2026). For 2026 to 27 that adjustment is shadow priced rather than applied, per IHACPA’s NEP26 HAC Technical Specifications (IHACPA, 2026), so the penalties are not currently applied to hospital funding. For how coded data flows into funding more broadly, see the guide to clinical coding and hospital funding.

One caution sits alongside Liu’s finding in the table above: Duke and colleagues’ 2022 chart review of reported hospital-acquired complication events at one Melbourne health service found no HAC coding or classification errors, although 25.5% of the reported HAC events were clinically false positives. Not every documentation problem, in other words, is a coding error.

Clinical documentation improvement (CDI) is the programme of work that tries to fix documentation before it turns into coding errors and audit findings. No national CDI standard, certification or programme was found in the research behind this page, so it is more useful to describe what documentation improvement actually involves in Australian hospitals.

In Australian practice, documentation improvement rests on mechanisms that already exist in the audit literature:

• Documentation queries from coders and auditors to treating clinicians, which resolve ambiguity at the source.
• Education for clinicians about what the classification needs from a record, as in Grealish and colleagues’ 2026 Sydney fracture study, where junior doctor education lifted correct M80 coding from 1 of 94 admissions to 14 of 39, though the improvement fell back within 10 weeks (Grealish and colleagues, 2026).
• Feedback from coding audits, where audit results tell coders and clinicians which documentation patterns are producing errors.

Published Queensland guidance asks auditors to give constructive, evidence-based feedback to coders and, where relevant, to clinicians about documentation, because coded data will not improve until feedback is given (QCAEC, 2017). Record-keeping has its own floor in the National Safety and Quality Health Service (NSQHS) Standards: Action 1.16 requires health service organisations to support the workforce to maintain accurate and complete healthcare records (ACSQHC, 2017). The Standards guide does not mention coding, so it should not be read as requiring coding audits.

Documentation improvement is also not a complete answer on its own: the Audit Office of NSW’s 2015 performance audit also names coder error among the common causes of miscoding (Audit Office of NSW, 2015).

For an individual coder, documentation quality shows up in three places: the queries you write, the audits you take part in and the feedback you act on. Across this series, a clinical coding audit means the independent re-coding of selected records and comparison with the codes originally assigned, per the Australian Institute of Health and Welfare’s (AIHW) 2026 admitted patient care data quality statement (AIHW, 2026). The Audit Office of NSW describes the auditor as a different clinical coder reviewing the codes assigned to an admission to identify miscoding, then determining the severity and primary cause of any error (Audit Office of NSW, 2015).

Within that process, a coder can act on documentation in three ways:

• Raise a documentation query to the treating clinician where the record is ambiguous or coder and auditor read it differently, rather than guessing (QCAEC, 2017).
• Use audit findings that record documentation as the cause of an error to direct feedback to the clinicians behind those notes, alongside feedback to coders (QCAEC, 2017).
• Take part in the feedback loop that follows an audit: constructive, evidence-based feedback to coders, with significant findings addressed in a structured education session (QCAEC, 2017).

Coding audit or clinical audit?

They are different disciplines: a coding audit checks the codes assigned to an admission against what the record documents, while a clinical audit checks the care delivered against clinical standards. For that other discipline, see the pages on clinical audit methodology in Australian healthcare and what audits in healthcare are, and keep the two terms apart in conversation as well as in writing.

Two audit guides in this series cover the ground in depth: what a clinical coding audit is and why Australian hospitals run them and the clinical coding audit process, step by step.

TalentMed (RTO 22151) does not offer a clinical coding audit course. Its HLT50321 Diploma of Clinical Coding includes taking part in coding audits of your own coding work in its final unit, HLTCCD007 Undertake moderately complex clinical coding. That participation is not training for a coding auditor role.