Data quality audit of a clinical quality registry: a generic framework and case study of the Australian and New Zealand Hip Fracture Registry

Aidan Christopher Tan; Elizabeth Armstrong; Jacqueline Close; Ian Andrew Harris

doi:10.1136/bmjoq-2018-000490

Data quality audit of a clinical quality registry: a generic framework and case study of the Australian and New Zealand Hip Fracture Registry

BMJ Open Qual. 2019 Jul 17;8(3):e000490. doi: 10.1136/bmjoq-2018-000490. eCollection 2019.

Authors

Aidan Christopher Tan¹, Elizabeth Armstrong^{2

3}, Jacqueline Close^{2

4}, Ian Andrew Harris^{5

6}

Affiliations

¹ Faculty of Medicine, University of New South Wales, Sydney, New South Wales, Australia.
² Falls, Balance and Injury Research Centre, Neuroscience Research Australia, Sydney, New South Wales, Australia.
³ School of Public Health and Community Medicine, University of New South Wales, Sydney, New South Wales, Australia.
⁴ Prince of Wales Clinical School, University of New South Wales, Sydney, New South Wales, Australia.
⁵ South Western Sydney Clinical School, University of New South Wales, Sydney, New South Wales, Australia.
⁶ Whitlam Orthopaedic Research Centre, Ingham Institute for Applied Medical Research, Sydney, New South Wales, Australia.

Abstract

Objectives: The value of a clinical quality registry is contingent on the quality of its data. This study aims to pilot methodology for data quality audits of the Australian and New Zealand Hip Fracture Registry, a clinical quality registry of hip fracture clinical care and secondary fracture prevention.

Methods: A data quality audit was performed by independently replicating the data collection and entry process for 163 randomly selected patient records from three contributing hospitals, and then comparing the replicated data set to the registry data set. Data agreement, as a proxy indicator of data accuracy, and data completeness were assessed.

Results: An overall data agreement of 82.3% and overall data completeness of 95.6% were found, reflecting a moderate level of data accuracy and a very high level of data completeness. Half of all data disagreements were caused by information discrepancies, a quarter by missing discrepancies and a quarter by time, date and number discrepancies. Transcription discrepancies only accounted for 1 in every 50 data disagreements. The sources of inaccurate and incomplete data have been identified with the intention of implementing data quality improvement.

Conclusions: Regular audits of data abstraction are necessary to improve data quality, assure data validity and reliability and guarantee the integrity and credibility of registry outputs. A generic framework and model for data quality audits of clinical quality registries is proposed, consisting of a three-step data abstraction audit, registry coverage audit and four-step data quality improvement process. Factors to consider for data abstraction audits include: central, remote or local implementation; single-stage or multistage random sampling; absolute, proportional, combination or alternative sample size calculation; data quality indicators; regular or ad hoc frequency; and qualitative assessment.

Keywords: audit and feedback; healthcare quality improvement; human error; quality improvement methodology; surgery.