Table 5 Data custodian and user perspectives on benefits and limitations of secondary use of primary care data

BENEFITS (themes arising)

Intrinsic benefits of primary care data

Unique, rich, granular, ‘real world’ data with capacity to provide more population health information than any other health data source. Makes regional and remote-level information more accessible. Minimises measurement bias in research. When linked it creates systems view and triangulation, creating a ‘patient centred view’ of care pathways, patient needs and service gaps.

Assists policy and planning for provision of improved health services and health outcomes

Its analysis enables greater knowledge to assess and improve services in localities or broadly, service quality improvement (through competitive benchmarking), understanding of treatment outcomes and population health improvements. Provides an evidence-base for investment, interventions and efficiencies in health spending and technical infrastructure. Can inform policy and workforce planning. Contributes to a ‘Learning Healthcare System’.

Pragmatic research efficiencies & improvements

Data driven research can be rapid and cost effective leading to cost reductions.

Big data (from large electronic medical record repositories) increases statistical power and increases research scale.

Big data research can illuminate aspects of primary care otherwise not seen and can generate new research questions.

Patient generated data

Technologies can facilitate patient reported data collection through add-on apps, with potential to enhance primary care data.

Practice level

Enables providers to review their activities and make business improvements and ability to track and improve patient outcomes.

LIMITATIONS (themes arising)

Technical and data capture limitations

Limitations from using or merging data from non-standardised clinical software systems and non-standardised data capture systems. Complex general practice workflows negatively impacting data quality.

Clinicians who use paper-based rather than digital records.

Poor data quality, reliability

Data captured in non-standardised clinical software systems and extracted using non-standardised data capture systems leading to data inaccuracies and loss of data context.

Poor use of existing field coding by clinicians and high use of free text limiting data utility (and adding to burden of data cleaning).

Incomplete data fields (fields not captured by clinician or not extracted for secondary use). Difference between terms used by GPs and available SNOMED Clinical Terms prompting free text entries, and preferred terms changing over time, limiting data standardisation.

Data governance and access requirements limiting its usability

Requirements that data be de-identified or aggregated limits its utility.

Lack of a minimum primary care/general practice data set in Australia containing data from all providers.

Lack of a nationally endorsed patient unique identifier limits data linkage and identification of patient duplication.

Limited permissions on what data can be linked and the ‘arduous’ nature of permissions process and cost to access data.

Little incentive for primary care providers to share data leading to limited data available for secondary use.

Poor understanding of data complexity and context

Many data end-users unable to appropriately interpret the data because they do not understand its social and clinical context.

Data collected for one purpose (clinical care) being used for purposes other than the primary purpose.

Unequal data representativeness

Lack of available data on priority populations including culturally and linguistically diverse, aboriginal communities, under-representation of vulnerable groups and over-representation of the ‘worried well’.

Privacy concerns, trust and ownership

Lack of community consultation on data use. Concern that shared data are stored off-shore or its use cannot be controlled.

Unclear consent mechanisms and privacy concerns not addressed limiting clinicians sharing data.

Varying ideas of who owns the data limiting the extent to which it is shared.

Lack of guidelines, policies, standards and ‘common data model’

The following limiting availability and utility of secondary primary care data use, lack of: national standards for general practice data quality and evaluation, clinical data capture system interoperability (too many clinical data systems), standardised data extraction tools, standard coding and common terminology; leadership to improve data standards and a ‘common data model’.