At its core, the medical record is a documentation system where clinicians record care and communicate assessments, findings, conclusions and plans to other carers. The record can also include orders and results, prescriptions and records of drug administration.
Traditionally, medical records in hospitals were separate from nursing records; however, increasingly these are combined. Computerisation provided the opportunity for integrating decision support, clinical guidelines and care pathways, data analytics and reporting, all of which promise greater efficiencies in workflow, and better outcomes for patients.
Computers replaced paper records in general practice quickly, whereas computerisation in hospitals was slower and largely driven by the requirement for data for disease registries and audit purposes, which led to many different electronic ‘clinical information systems’ that worked for one specialty, but not others, and the lack of standard data models led to major difficulties in exchanging information between them.
The single vendor hospital EMR solutions that we see today were largely a response to this lack of interoperability and have enabled some hospitals to become entirely computerised, more efficient and deliver higher quality care.
So why could this be a problem? After all, some of the most advanced and high quality health systems cite EMRs as being the major enabler of improvements.
Well, firstly the cost to license and implement remains high and only those with access to large amounts of capital and skilled resources can afford them. The cost of support and maintenance is also high and it is interesting to note the slow private sector uptake of these systems, particularly in Australia, where the business case for them is less clear.
Secondly, these systems tend to be built on technology that is rapidly superseded and lacks flexibility, and are consistently reported as difficult to use.
Finally, because these systems are bought and built primarily for hospitals, their scope restricts attempts to integrate care around the person throughout their care journey, which means that integrated care models that reduce the overall burden of healthcare to the population are difficult to implement.
There are more 'open' architected EMR solutions, and a number of open-source solutions which may be cheaper or even free to license, but these require a highly skilled workforce to tailor and maintain and they may lack sophistication and/or flexibility. Some organisations still pursue a 'best of breed' approach to their ICT, but this approach carries an increasingly costly integration overhead due to the variability in the underlying data models.
Standards for interoperability continue to mature, with HL7 and DICOM still having a role and FHIR being used more and more successfully. However, even if data is exchanged seamlessly between, say, a hospital EMR and an aged-care system, the ability for either to process that data at the enterprise level to drive true integration of care is limited. We still need something that sits across the entire health system so that care can be truly patient-centric and provided more cost-effectively closer to home.
Many would argue that electronic health records (such as the My Health Record in Australia) bridge the gap between EMRs by providing a longitudinal view health events for the patient. However, as long as the focus is on capturing referrals and discharge summaries (which reflect non-integrated models of care), and structured data is limited to specific pre-defined items, then decision support running across organisational boundaries will be at best unreliable.
The technologies that drive internet shopping and mobile applications offer a more flexible solution. The advent of ‘data lakes’ and cloud computing, which enable access to data via open APIs and has facilitated the explosion of innovative mobile applications, are equally applicable to the healthcare context. Data are supplied by systems that run different areas of the health business (these could be EMRs but notably don’t need to be) and is supplemented by the data provided by use of apps.
Algorithms can work across the enterprise, and beyond, to support decisions and drive workflow through notification tools, independent of the organisational constructs or systems behind them. A solid understanding of clinical data semantics is still essential, so processing of data is still required to convert data and terminology to a common form, however judicious selection of source system reduce the risk of complexity, avoid competing data models and terminologies, and ensure that APIs are accessible. This is the future of the Open EMR.
Several large healthcare providers are already looking at data lakes with Open APIs to drive value and flexibility into their business, and enable more patient-centred care. Some have EMRs in situ as well but others, like Calvary, are looking to open data lake technology to support cost-effective integrated person-centric care across hospital, aged and community care settings without the expense and constraints of the single vendor EMR.
Dr Robin Mann is the National Chief of Innovation at Calvary Care