A clinical workforce that understands and believes in the value of data will engage and take care in entering relevant and accurate data.
Mistrust in data quality may stem from not having visibility of the processed followed to collate, analyse and check the data. Clear communication on steps followed, including any limitations in the data will help provide transparency and inform clinicians on what they are working with. The reality is that data quality improvement is an iterative process, and it is difficult to aim for ‘perfect’ data from the start. A more practical approach would be to clearly communicate current situation and processes as above, and then to make data available to clinicians and work with them to identify opportunities to improve data. Apart from assessing systems in use in the organisation (and the related user experience in using these to input data), other potential opportunities could be in identifying the data that is currently being inputted to prioritise subsets to focus quality improvement initiatives on and to identify opportunities to automate data input to free up clinician’s time to focus on priority subsets.
Technologies such as EPRs, used in the right way, can improve data quality and patient outcomes.
However, poor usability of EPRs can burden already stretched frontline staff and take up valuable clinical time. A survey done in 2019 of EPRs in the Emergency Department suggested that none of the 25 systems included met the acceptable usability standards.
Dr Peter Thomas, CCIO and Director of Digital Medicine, Moorfields Eye Hospital believes that headspace and time in which to enter structured data are key issues to high data quality rather than data literacy. He highlights that clinical systems can be difficult to use and can involve a lot of ‘clicks’. Peter explains that this adds to time to already stretched clinical consultations, and that while structured data helps the next consultation and ability to gain better insights from the data, time pressure leads to clinicians finding workarounds including entering structured data into unstructured fields such as ‘patient history’ free text boxes.
Peter notes that Large Language Models (LLMs), such as ChatGPT, can be quite successful at identifying clinical concepts like a primary diagnosis from within free text. He believes that this will allow clinicians to do what they do well: record consultations through narrative text. AI technologies can then extract structured data items for entry into EPRs. Such an approach could substantially reduce the amount of clinician time required to record a structured clinical note
As an addition to what Peter describes above, AI technologies that provide ability to translate speech (i.e. dictations) into written text can automate the text entry first step further making the process easier for clinicians.
Data literacy is essential. Training on data (and statistics principles) as applied to healthcare should be embedded into undergraduate curriculums and made available to the existing workforce to complement ‘on the job’ training. The training should include how new technologies and data analysis techniques can support research and evaluation of the health economics of interventions. A learning framework which develops and maintains data literacy and capability is helpful and this needs to be supplemented by developing communities and encouraging champions that will support others and promote data literacy. As advanced technologies such as Artificial Intelligence becomes more prevalent and has the potential to improve patient outcomes and healthcare operations, training on these technologies also must be considered. This will create a more confident clinical workforce when it comes to data and technology and will support the informed and safe use of data and AI solutions, where the workforce understands both the potential and limitations, and is able to derive insights appropriately and work effectively alongside technologies. Investment should be made in creating hybrid clinical roles with dedicated time to data and health technology solution development and implementation which could also help engage and retain a clinical workforce by providing variety and a means to pursue their areas of interest without leaving the NHS.
