In this post I want to explore the impact of poor data quality on the new Summary Care Record database being implemented by the UK National Health Service.
In particular I want to discuss how poor data quality leads to a cycle of increasing mistrust and gradual data quality degradation.
Trust and the Negative Data Quality Feedback Loop
First, some background. The UK National Health Service (NHS) wants over 50 million people in England to agree to their records being included in an individual Summary Care Record (SCR) database. The aim of this program is to help improve the quality of care provided by hospital staff and any out-of-hours doctors by releasing information that is typically held only by a person’s general practitioner (GP).
Following a pilot project in Birmingham it was revealed that 1 in 10 patient records loaded onto the national Summary Care Record system had not been correctly updated, leading to concerns from health experts that fatalities could result from applying inappropriate drugs or treatments.
This story is all the more significant in light of the recent case of a German out of hours doctor who administered an incorrect dose of a drug to a patient which caused fatality. Although the doctor in question did not use the Summary Care Record system the results of what can go wrong are all too apparent.
The results of the recent pilot have prompted the British Medical Association (BMA) GP Committee to pass a motion suspending the database until the safety issues have been fully investigated and resolved.
With 10% error rates being cited it would appear that any resolutions would take considerable time and effort, if indeed they are at all feasible.
NHS managers apparently claimed that “…the problems were insignificant because all care records came with a ‘health warning’ that they should not be relied on”.
This begs the question of trust.
If the perceived trust of the data is so low – how will the system function with a live, national rollout?
Why is a Lack of Trust So Damaging?
A lack of trust does not directly create poor data quality, it is the events surrounding a lack of trust that ultimately generate poor data quality.
For example, a utilities engineer goes to install a new piece of equipment. The wiring schematic from the CAD based system is found to be inaccurate. New equipment is missing and the power rating is incorrect.
If the trust in the system is strong then the engineer is likely to maintain the system correctly but if the trust is low then the engineer is likely to store some basic information in the system and then perhaps add their own personal records using their own local spreadsheet.
On every data migration project I’ve been involved with, local records were discovered and one of the overriding causes was always because the workforce did not trust the system to store and retrieve the information they required in a format that was right for their role.
As a result the data quality worsens, caught in a continuously degrading feedback loop.
What is the answer?
In the example of the NHS database, trust is critical to the success of the system. People have the right to opt-out of the system and if the press surrounding the quality of data in the system continues to be negative then allergy sufferers and people on specialised drugs for example may certainly question their involvement in the scheme.
The answer is an obvious one.
Don’t try and resolve data quality once the system is live, it is simply too late. “Fixing the data in the target” is a classic data migration tactic but one fraught with risk because although we may eventually be able to recover data quality levels it can take far longer to recover trust.
The main focus of the project leaders would appear to be completion first, quality second.
If the reports of 10% defect rates in patient records are accurate then the entire integrity of the system is in jeopardy. If that figure does not dramatically improve there may be no way to retain and grow public trust in the system moving forward.
Which is truly sad because the system, as a concept, presents an excellent example of how data of the highest quality can save lives.
- NHS unique ID failure leads to duplicate newborn patient records (Data Quality Pro)
- Patients’ medical records go online without consent (The Telegraph)
- NHS database raises privacy fears, say doctors (The Guardian)
- ‘Opt out’ fear on electronic medical record system (Times Online)
- Firm that hired fatal overdose doctor Daniel Ubani hopes for takeover (The Guardian)