The rubber has now hit the road as far as rolling out the UK’s Summary Care Records System (SCR) and we are now seeing the inevitable and predicted difficulties.
The following provides a flavour of what is going on
Confidential report finds serious errors in new central patient database
Tony Collins
Friday 19 March 2010 09:45
The Summary Care Records database - which is central to the government's plans to create health records for 50 million people - contains inaccuracies and omissions that make it difficult for doctors to trust it as a single source of truth, according to a confidential draft report.
The findings by researchers at University College London, are likely to reinforce the concerns of the British Medical Association which has called for a halt to the "rushed" rollout of the "imperfect" Summary Care Record scheme.
The Government launched Summary Care Records to help doctors and nurses make better clinical decisions. The aim is for clinicians and out-of-hours doctors to have access, particularly in an emergency, to a central record of a patient's allergies, medications and adverse reactions to drugs.
But the researchers at University College, London, found examples where the Summary Care Records central database failed to indicate a patient's allergies or adverse reactions to drugs, and listed "current" medication that the patient wasn't taking.
The database also indicated allergies or adverse reactions to drugs the patient did not have.
Inaccurate information in the central database came from uploads of patient records by GP practices.
Patients unharmed - because doctors didn't trust the SCR database
Researchers at UCL found no evidence that incomplete or inaccurate data on the SCR database had led to patients to coming to harm - but precisely because doctors did not trust the new system as a single source of truth, and took extra time to double-check details of medications and allergies.
No evidence of safer care
The researchers found no direct evidence that the care records system led to safer care, though they said that access to the database may reduce some rare medication errors. There was no clear evidence that consultations between doctors and patients are quicker - and in some cases use of summary care records made consultations longer.
But researchers also found that when central records are accurate, they can be useful for clinicians, particularly when patients are poor at communicating or, if they are on multiple medications, cannot remember what they are.
More here:
Additional links
Confidential report on Summary Care Records finds database is inaccurate - IT Projects Blog
New patient records database contains errors - The Times
We also have
GPs join attack on SCR roll-out
23 Mar 2010
GP leaders have joined the attack on the roll-out of the Summary Care Record, calling for an immediate halt to the roll-out and an urgent discussion of the issues with the Department of Health and NHS Connecting for Health.
The BMA’s General Practitioner Committee, which represents all GPs in the UK, passed a unanimous motion saying that "the GPC deplores the recent fast roll-out of the SCR in England.
"We seek the halting of this roll-out and that the DH and CfH discuss these issues urgently with the profession.”
Dr Grant Ingrams, co-chair of the joint IT Committee of the BMA and RCGP, said the general feeling of the GPC was that the roll-out had been “a dog’s dinner."
He told EHI Primary Care: “It’s been pushed through without thinking through the consequences, there is a lot of misinformation out there and there are a lot of primary care trusts and certainly practices who are not really up to speed on the issues.”
The debate at the March meeting of the GPC follows a letter sent by the BMA to health minister Mike O’Brien almost two weeks ago, calling for the SCR roll-out to be suspended.
More here:
http://www.ehiprimarycare.com/news/5759/gps_join_attack_on_scr_roll-out
and here:
SCR suffers from variable GP data
19 Mar 2010
A confidential draft report from the evaluation of the Summary Care Record says data uploaded from GP practices is sometimes wrong but that the SCR can be useful when the data is accurate, Computer Weekly magazine has reported.
The IT magazine says the evaluation team from University College London found examples of uploads from GP records where there were inaccuracies and omissions in the data on medications, allergies and adverse reactions.
The researchers found no evidence that patients had come to harm because of the inaccurate or incomplete data but said doctors took extra time to double check details of medications and allergies, according to the magazine.
Concerns about the quality of the data uploaded from GP practices has been a theme since the SCR was first devised with GP practices originally incentivised to provide good quality data through an IM&T directed enhanced service which sought to create data ‘fit for sharing’.
Since the DES ended in March last year the quality of the SCR has been reliant on primary care trusts implementing data quality standards which has been criticised by some as providing no “carrot or stick” to practices on data quality.
The draft report from UCL is also reported to say that SCRs can be useful for clinicians if they are accurate, particularly if patients are poor at communicating or if they are on multiple medications and have difficulty remembering them.
Other findings reported are that there was no evidence that the SCR made consultations shorter and that in some cases it made consultations longer. There was also no direct evidence that SCRs led to safer carer although it might reduce some rare medication errors.
More here:
http://www.ehiprimarycare.com/news/5755/scr_suffers_from_variable_gp_data
The lessons for NEHTA and Australia from all this are legion.
First it is just silly to roll out this sort of thing until you have the quality of the basic patient data in GP and specialist systems really up to scratch. That of itself is a 5-10 year project. The UK had been providing large data quality incentives for a number of years under the banner of ‘data fit to share’ and the data is still flawed. Can you really believe the situation would be even as good here?
Second, rushing implementation of this sort of top down approach will inevitably take time. There is no sense in pushing too hard. Incremental steps are vital.
Third is that the UK has a limited number of different GP systems that have been planned over years to capture the data set required. We are at present nowhere near having that level of standardisation and to get there will take a long time.
Fourth it is clear the evidence supporting the value of a centrally stored shared record is presently not all strong or compelling despite the intuitive view that it is a great idea.
Last there are all the issues identified in the blog of a couple of days ago about consent, information sharing and control, physician trust and so on.
See here:
http://aushealthit.blogspot.com/2010/03/serial-commenter-who-has-something.html
Read closely about the UK lessons so far. This is a path we may not want to sensibly go down without a great deal of thought and care!
David.
At the risk of shameless self promotion, here is a paper I wrote for HIC'08 on some of these issues.
ReplyDeleteIn particular, the data quality is one that I believe can only be solved by someone being responsible for the organisation/maintenance of the information (and this is fundamentally not compatible with a 'everyone contributes equally' central shared record)
The latter half of the paper may be some food for thought, especially with yesterdays announcement of GP's being paid to be 'responsible' for signed up diabetics
(presumably they should be responsible for
the shared records of their diabetics as well then??)
http://www.patto.net/papers/caseagainstsehr-patterson.pdf
Don't take the suggestions in the paper too seriously - I'm pretty sure I don't have the actual solution to these very difficult problems, but I'm equally sure that central records aren't the answer either!
I re-read Patto's HIC 2008 dissertation and recognised the intellect of that contribution to eHealth thinking. Whether the ultimate SEHR solution is a centralised or federated one I feel that the pathway to a federated solution, which has some attractive features, is a pretty tortuous one. I notice Patto referred to the use of the Medicare number as a link and see the transition to unique health identifiers alone as a challenge as it is taking a while for NEHTA to weave its magic in this area.
ReplyDeleteAnyway back to the subject of the blog... Government support for all clinical providers to acquire systems that help to drive data quality improvement at source would be helpful. That will assist federated or centralised SEHR solutions and will add to quality and safety in care provision for patients as we move along. While there are some Government drivers helping this to occur, they are tiny insignificant when compared, on a per capita basis, to the investments made by the UK NHS.
I have to be careful here as I also may be as guilty as Patto of promoting a strategy that is linked to a company I know.
John Johnston, I see nothing to be ashamed of if you are promoting something that is of value and can provide benefit to practitioners and consumers, like the work you are doing. In that spirit, and with a similar qualification, my own 2 cents worth:
ReplyDeleteWhile I agree that the path towards federation is difficult, I am working to deploy an online health IT service that is "federation ready", and is already aggregating information from many sources. It provides a service to support GPs, patients and their care teams in chronic disease management. The aggregation of information is focused on that "slice" of the patients' information relating to the management of their chronic disease.
Health care providers are willing to contribute information to this record because it provides value in return. Consumers are willing to consent to their information being shared because they appreciate the value of this. Other aspects of the design of this service contribute to a "positive feedback" loop that will tend to gradually improve the quality of data in the overall system and in the systems of each contributor at source. Not least of these is the fact that data is being used and exposed to multiple eyes (including the patient) and used to support collaborative care.
I think systems such as John's and this one are worthy of support because they do point the way towards the kind of "bottom up" solutions that are required to really make health IT work, especially in primary care.
Jon, the point I was making in my paper was that whilst it is certainly possible to have a federated architecture that aggregates from many sources, the most privacy neutral architecture is one where the federation 'nodes' are the actual GP clinics and clinical data is only ever held in the systems of the clinicians for that patient - it isn't even aggregated out to other third party systems - the sharing takes place IN the GP system (i.e the GP would install your chronic disease system on their servers).
ReplyDeleteSo whilst I understand the IHE/XDS federation model has some improvements over centralised, I'm not sure how it addresses some of the problems. For instance, there is still an important data quality issue - everyone is responsible for the pieces of information that contribute to the aggregate, but noone is responsible for the overall picture of the aggregate. Noone is responsible for asking the patient about the continued relevance of certain information (perhaps they had chest pains whilst they were on holidays - and saw a GP - who follows up on that visit? Who corrects/clarifies the other GP's preliminary diagnosis?).
As I say in my article, all these aggregation models seem to me to be like everyone throwing pieces of paper into a filing cabinet and assuming they will sort themselves. One aspect of data quality is making sure what's written on the paper is correct. But another aspect is to make sure the collection of paper tells a coherent 'story'. I don't see how pure aggregation systems do that?
Hi Andrew,
ReplyDeleteI understand the point you are making, and I agree that if the system is "purely" aggregating data and no-one is using it then there is no driver towards data quality.
If on the other hand, people are using the data then is is possible to design systems in such a way that data quality is gradually improved.
The system I describe is not "purely" aggregating data - it is aggregating data as part of a planned care management process that is under control of a care coordinator (GP in this case.) This means that, with careful design, data quality can be gradually improved because it is constantly under review. This is the basis for the claims in my previous post. Improving data quality in source systems is less direct. As errors are found in the repository that relate to errors in the source data, we have put in place systems that encourage users of the shared system to correct the data in their source system and resubmit to the shared space, replacing the incorrect data.
The important thing here is that the repository is not "purely" aggregating data - the data is in active use, and systems are designed to encourage (and reward) correction of errors in the data.
While the current repository is not, in fact, an XDS repository, I can see no reason why the kinds of workflows I describe cannot be implemented using an XDS repository as a data store. To see if this can be practically realised, I am working to develop an IHE profile for a "Patient Centred Coordination Plan" that attempts to put into practice some of the things I have learned from the implementation of the "stand alone" web based system I briefly described above.
I must be doing something right, because I have attracted significant attention (and occasional dispute) from both the clinical experts in care planning and the technical experts in workflow within IHE International. Quite an experience, and the journey continues...
By the way, if you are interested, the work in progress can be viewed at http://ihe-australia.wikispaces.com/Care+Coordination+and+eReferra
Assistance in developing and documenting the concepts within will be gratefully accepted.