i-HD’s Dipak Kalra examines the importance of data quality to eSource-readiness in hospitals

esource readiness

i-HD’s Dipak Kalra examines the importance of data quality to eSource-readiness in hospitals

What is eSource?

‘eSource’ is understood well by some in our health and research ecosystem, and is an entirely opaque, mysterious word to others…

The term was coined by the medicines regulators, to reflect change in the way clinical trial information could be brought into a study. There’s been increasing interest over a number of years in reusing data that might already be captured during healthcare provision, which, provided it meets certain requirements, would be entirely valid to include in a clinical trial. eSource (or electronic source) covers any of this information that may already be available in digital form, rather than needing to be entered manually into clinical trial systems.

One form of eSource data that is going to explode (or is already exploding) comes from the use of digital innovations like apps and wearables. The opportunities to use this data, captured directly from patients, in clinical trials, are significant. Today, however, the primary interest across the research industry is in making better use of the data clinicians enter into electronic health records (EHRs).

Transformational technology

What we are now seeing is a very sharp and accurate mechanism for using EHR data as eSource, through the introduction of a new set of players in the ICT marketplace, who act as a kind of bridge between the EHRs and the clinical trial systems in use in pharma or academia. This enables those who are designing studies, recruiting into studies, and then conducting studies, to be able to leverage the data in EHRs.

Those conducting trials can design their trial in a way that’s likely to succeed; know which sites have relevant patients; help to recruit those patients by case-finding through querying the EHR; and then move some of the relevant pre-existing eSource clinical data directly into the clinical trial ecosystem. So, investigators don’t need to sit there at a screen, re-typing that information, with inherent errors, time and other factors.

In this way, we’re seeing a really exciting reuse of EHR data, right across the lifecycle of clinical trials.

The importance of quality

One of the things that all this means, is that the quality of data entered into EHRs is more important than ever. Clinical trial data has to meet certain quality standards, and so, when you bring in data that is sourced from a real-world environment – for example from a hospital or patient – it has to meet those same standards for regulatory decision making.

What does good quality data look like? Over the last couple of years, guidance documents have been published by the medicines regulators, particularly the FDA (in the US) and the EMA (in Europe), that outline their expectations for eSource data to be acceptable in clinical trials. In my career, I have also developed a number of international standards, through ISO, for the qualities of good electronic health records for clinical care purposes.

What I’ve been pleasantly surprised by, is that the eSource requirements from regulators and the requirements that myself and colleagues have established for good electronic health records, have been very closely aligned.

So, if we want to use clinical information in an EHR, there are things that a good system – and a good data-creating organisation – needs to take care of. These requirements apply irrespective of whether the goal is to reuse the data for research or for any other use, and – ultimately – they are about being able to trust the information. Just a few examples of properties that you need to have full confidence in to be able to trust clinical data include:
● who the patient is
● when the record was created, and by whom
● the version history, detailing if information has been modified over time, for instance to correct any errors.

In addition, when we want to re-use the data outside of the originating organisation – for instance for a clinical trial – we need evidence in support of the information.

Another aspect of data quality that’s of particular relevance when re-using data, is in the detail of the data itself. For example, where a diagnosis has been entered using a coded terminology system, we need to know where the coded terminology comes from, and which version was in use at the time the data was entered. If a measurement has been entered, what are the units of measurement?

Some of these considerations sound obvious but, unfortunately, there are not always robust systems in place to ensure that high quality, re-usable data is recorded.

High quality data: the challenge

One of the reasons data quality isn’t better than it is, is that, unfortunately, we have a very strange return-on-investment ecosystem around health data today. Many clinicians are well used to working with poor quality clinical information, in tatty paper notes or blocks of free text on a word processing document. Continuity of clinical care goes on fairly well, on the basis of really quite poor amounts of structured and coded computable data.

In essence, the people who stand to gain most from better quality data are not the people that enter it. If you think of the tired junior doctor or nurse at 3am, there is no motivation for them to take the time to really structure and code the entries on their patient notes, and to detail all the negative and positive findings.

And yet, the pharma industry, the med tech sector, public health agencies, health insurance, health ministries, and a number of other bodies, are desperate for high quality data. Those of us that are invested in data quality are encouraging our industry colleagues – those that stand to gain most from better healthcare data – to be part of the solution. To see a genuine shift in quality, those colleagues must become co-investors in the data quality improvement environment. And, in fairness to them, they are always very receptive to the idea that this is an industry-wide challenge that needs a multi-stakeholder approach.

How can hospitals, specifically, improve data quality?

Despite these system-wide challenges, there are still practical steps that hospitals can consider, when looking to improve data quality. Of course, the simple answer is to say that, if we had more doctors, nurses, physiotherapists, pharmacists, etc, clinicians would all have lower caseloads, and more time to spend on documentation. But we all know we need to be more creative than that. More realistic practical steps can include:
● Upgrades to technology connected to EHR systems – speeding up data entry by using voice recognition, touchscreens and better user interfaces, will ensure that entering better data doesn’t need to take longer than it used to.
● Training – so that everyone responsible for entering data is working to the same set of requirements and has the relevant knowledge.
● Feedback loops – such as data quality dashboards that introduce visibility and accountability for data quality.
● Appointing a data quality manager – who will act as a champion for data quality and be responsible for looking at what works and doesn’t within the organisation is, anecdotally, driving real improvements in some hospitals.
● Ensuring better data leads to better patient care – through intelligent guidance and alerts that allow clinicians – and, ultimately, patients – to benefit today from the data that was entered yesterday.

The last point is perhaps the most important of all – if patients start to see the benefit from computable data, and clinicians can see that benefit to the patients they care for, then they will, inevitably, care more about the data they enter.

Getting started

The above are all practical solutions, though they do still require varying levels of investment and effort in the medium term. For those just getting started, there are easy ways of dipping your toe into data quality improvement that we, at i~HD, can help with…

● A data quality assessment – can be run by us on a data extract, giving an assessment of how data stacks up against the data quality dimensions we have designed, and what the issues are that may need to be addressed.
● A free self-assessment survey – is an even simpler way to take a snapshot of where you are, without the need to gain approvals for a data extract.
● Peer support – In the near future, i~HD hope to be able to bring hospitals that are keen on improving data quality together into a Network of Excellence, for peer support, knowledge sharing and expert advice. Watch this space!
Moving forward
My call to action on data quality is always a multi-stakeholder one, because, as I’ve already highlighted, those that use healthcare data need to invest in the system that creates the data.

To hospitals in particular though, I would say that there are many reasons to focus efforts on data quality, and most of these are not related to clinical trials. Data quality impacts on the safety of patients’ care; it impacts on the effectiveness of your care; it can help you with managing costs; it should help you with improving outcomes; it may also help you with internal research on improving care pathways.

For all these reasons, I would encourage hospitals that know they have data quality work to do, to take those first steps in an improvement journey. Beyond the initial stages, you will need to be prepared to start allocating a budget, because data quality doesn’t go up for free. It does need to be made a strategic priority, and it does need investment. But you will be rewarded for that investment.

***

Dipak Kalra is President of the European Institute for Innovation through Health Data (i~HD), and is Professor of Health Informatics at University College London and Visiting Professor of Health Informatics at the University of Gent. He plays a leading international role in research and development of electronic health records (EHRs), including the requirements and models to ensure the robust long-term preservation of clinical meaning and protection of privacy. He leads the development of ISO standards on EHR interoperability, personal health records, and EHR requirements, and has contributed to several EHR security and confidentiality standards.