
There are varying definitions of data governance, but I think of it as a set of transparent rules that govern the quality of the data collected, how the data will be used and by whom, how the data will be kept secure and also compliant with the relevant legislation for the data collected, including Privacy legislation. This set of rules is critical for data collected by government in particular, where the community rightfully expects that data will be collected lawfully, held and used securely, for optimal advantage of the population.
Data quality
Data governance sets the standards by which the data will be collected. Standardisation of the data is critical to supporting high quality analysis and use of data. For example, measuring the same item in the same way over time means that if the data suggest that there is a change in trend over time, we can be more confident that this reflects a real change in trend rather than differences in how the data were managed over time. Equally, standard data support analysis of data, reducing the chance that bias or error can interfere with interpretation of findings. Standardisation across different data systems allows comparison with others, or as appropriate, pooling of data where greater numbers are needed for analysis. For example, state based data that are reported to national systems is supported by data standardisation across Australia. This standardisation is achieved through having standard definitions and defined values of what is to be reported. The staff collecting the data are trained in data coding, use a coding manual which specifies how the data will be extracted and recorded, and there is a data dictionary which explains to all users what the data represent.
It is particularly important to have a high degree of data governance in the setting of collection of identified data. Identified data in a formal system supports the key tasks of improving the quality and ascertainment of data through enabling linkage to other data collections. For example, linkage with another more complete data collection identifies if any relevant cases have been missed so they can be added on. Also, the data quality can be improved by using the linked data to improve, for example, identification of Aboriginal and Torres Strait Isander people in the collection, where it might be missing or incorrect. Identified data, through enabling data linkage, can support research and also limit the amount of data that need to be collected (efficient) to only the essential data required for the specified purpose. For example, if we were to have a research question relating to Veterans and death by suicide, linkage between Department of Veteran’s Affairs and the SA Suicide Registry would provide a complete data set to examine the research question. The alternate option of collecting veteran status in the Suicide Registry, would be incomplete as the registry relies on using existing data collections as the source, which may be missing critical information such as veteran status. If the purpose of the registry is to support the delivery of services, then identified data are necessary to be collected.
How the data are used
Data use options depend entirely on the purpose of the system, and can range from research, service and policy planning, through to supporting service provision. Further detail is provided below regarding these potential use cases.
Service support: a data system can both be a repository for information and a way of recording service actions. It can track end to end care, identifying where people may have fallen through the gaps and triggering the need for services. Service data also can be used to identify safety and quality of service provision. Systems should have an evidence-based standard for what high care means. This can be operationalised in different ways, but for example protocol based or guideline based care can be used. This is often characterised as guideline or protocol adherent care (or not), or differences in outcomes. These differences in care or outcome can be examined at the individual service provider level, at the organisational level and across the state. Analysis identifies care within agreed benchmarks and also outlier performance.
Planning: data can be used to support commissioning or service provision or guide the need for policy/legislative reform. For example, if the system collects service and outcome information, identifying areas of the state where there are lesser services/outcomes for the community, would guide the targeted commissioning of services to the gap.
Research: high quality systems provide data for research purposes, which is an opportunity for a deeper examination of data by relevant experts to understand new trends, issues that are amenable to intervention and at times, support the evaluation of trial outcomes. This ongoing research is critical to a high functioning system, identifying what is working and what is not, building the evidence base for best practice over time.
Evaluation: any intervention in the system, including different services, way of providing services, distribution of services can be evaluated by a well designed system. These systems generally would include service actions in addition to outcomes for families to be most useful from an evaluation perspective.
Data release
The rules for data release must be made clear from the outset. These rules can be further defined by legislation, but all data collected must be managed within the broader legal environment for protection of privacy. Further to legislation or regulation, rules for how the data will be managed within the collection itself should be defined, generally in a data release guideline. Data systems must clearly define these from the outset to prevent both inappropriate data release but also inappropriate barriers to the use of data. Reporting of analysed data should occur publicly, so that a range of stakeholders have access to the information. Registry systems often publish annual reports, ad hoc analyses on topics that have been examined in more detail, and more commonly now also open access data (as appropriate to the data itself).
A note about Aboriginal data governance: There is a large and growing expertise in Aboriginal Data Governance in Australia. I would encourage learning about this directly from key Aboriginal and Torres Strait Islander organisations such as the Lowitja Institute (328550_data-governance-and-sovereignty.pdf (lowitja.org.au)).
コメント