Related to that, effective information exchange, one of the key elements of successful and seamless care co-ordination, is often lacking. During our research project, we worked with five case study sites that are delivering effective care co-ordination programmes. We found that the sites used a ‘high-touch low-tech’ approach to information exchange – with face-to-face or telephone conversations, and secure email and fax. While practitioners especially enjoyed the personal contact with each other, there was also an overarching sense of frustration at the lack of access to shared electronic records, and having to update patient records on different systems was often cumbersome and time-consuming.
The common link between impact measurement and information exchange is, of course, information technology: data collection and care co-ordination involving multiple providers both depend on compatible IT systems. Health and social care have a fragmented IT framework, with as many systems as there are entities. Social care, primary care, mental health care, community care and secondary care all use different software, leading to a whole host of problems. These include: duplication of data entry with associated risk of error; differences in the indicators collected making comparison difficult; and incompatibility of data formats which complicate data exchange. Methods and systems to measure evidence of the impact of care co-ordination and integration are mostly weak, and policy-makers and commissioners are slowly waking up to this problem.
Practitioners in our study also highlighted the fact that software could not easily be adapted to reflect local circumstances; a key tenet of integrated care is that it operates best at a neighbourhood level and should therefore use indicators that reflect local demography, health needs, circumstances, and socio-economic conditions. Most existing software appears not to be flexible enough to meet these needs.
Commercial software is often difficult to adapt, as the source code cannot be manipulated, and, in some instances, the software can limit the amount of data that can be exported to other formats, complicating information exchange and the tracking of patient status. Some software carrying out predictive risk modelling, which uses patient information from the past to assess the risk of future hospital admission, will not allow users to access details on how indicators are weighted, so that providers cannot be sure if they are targeting the right patients. Furthermore, providers often have contracts that tie them to particular software and so may not be able to switch easily.
As a result of these limitations, there appears to be an appetite for developing in-house systems, using open source software that can be transferred to providers at little or no cost. Open source software allows anyone to access the source code and to adapt it to their own needs. This is, of course, not without its problems – if too many people in one organisation change the code, it will become difficult to track and understand changes. Changes must therefore be logged meticulously. Nevertheless, data analysts and practitioners appear to value the advantages of local adaptability and transferability, and national policy appears to be moving into the same direction, with the adoption of open source software for centrally provided NHS IT services.
Commissioners need to decide which system will best suit their local care-co-ordination needs. Whichever they choose, they need to make a conscious decision early on in the process – everything from patient identification and care co-ordination to future up-scaling of the model is simplified if suitable IT systems are in place.
Open source software will not, by itself make it easier to integrate social care, primary care, mental health care, community care and secondary care systems. The adoption of open source software cited is a national organisation where the scale means it is economic to employ a sufficient critical mass of developers to ensure a sustainable offering. Most NHS organisations are of a much smaller scale and will not be running software development teams to enable the systems to be amended - the Royal Marsden, a highly specialist hospital, is a rare exception where the software required was simply not available on the open market.
For most organisations their systems will be proprietary and therefore expensive to develop interfaces for.
The other complication is the range of organisations required - social care, primary care, mental health care, community care and secondary care all interact with each other, and in many cases multiple providers. The overlapping boundaries of the catchment areas (especially in London) make it very difficult to identify priorities for such integration, especially when the NHS organisations involved are subject to repeating reorganisations. This facet led to the development of a summary care record, only for this to be subject to concerns around data protection when operating on a very large scale. And in any event, the summary care record often does not provide the sort of detail that a team responsible for the shared care of a patient need to provide continuity of care.
Integration is especially challenging for such circumstances, where data entered once must be available to all - but only those who are directly involved in the current episode of care. Interfaces are generally easier to organisation on a push basis rather than pull - it is easier to download data from a system than to accept it from another. Reasons include 1) the difficulty of matching data - you may know me as Howell Huws, but that system has me down as Hywel Huws - even with the wonderful NHS Number and 2) the difficulty of matching data structures - my system may not have anywhere to put your specialised dataset. And then you get into how do you provide assurance of data integrity when data items may be changed by people on different systems simultaneously - whose is the master record? Hence the complexity and hence the expense of interfacing between the systems, and why it is so often put into the too difficult pile, and why so few actually get started.
An individual transaction-based approach (referrals, results, discharge notification) is likely to be easier to match to data protection issues, but does not resolve the shared input issue. Perhaps what is needed is a shared case record that only exists for the lifetime of the episode of care, contributed to by all organisations from data entered into their individual system and is then deleted when the GP decides (with the patient's consent, hopefully), that the episode is complete?
It's great to see one of the key challenges in Integrated Care finally being acknowledged; not great to see a complete misunderstanding of commercial software and open source expressed again.
The problem is that organisations have chosen the wrong kind of solution. If you buy a cheap "black box", that ticked a form for "risk stratification", you can't then complain that the model is not open or adaptable. But then trying to build the software from scratch in open source is not the answer either - most NHS Trusts just don't have the resource.
The problems of building personalised, localised models of risk, and bringing together complex sets of data to build these models as identified by the commentor above, have been solved in other countries and in other industries. From Tesco Clubcard, to Kaiser Permanente, this is an understood problem with standard tools and approaches to solving it. The real challenge is that the NHS has yet to really understand how key this element is to ultimate success - although hopefully this blog is the start of that understanding. Buying a risk strat tool off the shelf is not the answer. And the start of that answer isn't as hard as some people think.
I know very little about IT solutions and so have no axe to grind. I was being laughed at the other day by my junior doctors for mistaking a patient's I pad for a kindle and his ipod for a smartphone. But.....I will pose the questions of the naïve ignoramus. Whenever public sector IT contracts go belly up and the software and platforms are unfit for purpose (whether NHS, Whitehall or beyond) why is it always the public sector employees/government officials who are blamed for their naivety and lack of expertise in negotiating contracts with the private sector and not the companies/consultants who walk away, having failed to deliver with seemingly few consequences. Are they not equally culpable? And how "ethical" is a business that takes big money from cash strapped public services for a product that isn't fit for purpose.
Second, I know that the current DH has made better informatics and a paperless NHS and person held transportable records a priority and key to the integration agenda and of course I agree with this. But is it a triumph of hope over experience when the last NHS IT project cost £8bn (the entire primary care prescribing or criminal justice budget) for what? How can we ensure its different this time. A triumph of hope over experience, driven by the fact that generally conservative ministers are more savvy when dealing with business/industry?
In almost all walks of life beyond healthcare, the way we collect and use data has changed beyond recognition, creating enormous value in the process. Our consumption of media, is just one example of how clear incentives and ease of adoption have enabled rapid change.
Healthcare has not yet embraced this opportunity. Perhaps the fear of getting it wrong, or the lack of experience of organisational change or indeed the desire to preserve the status quo in the face of limited resources and increased demand have all had a part to play.
Meanwhile, as a patient I want to see providers use my data to deliver services more efficiently and effectively yet am astounded at how slow they are to grasp the opportunity.
Commissioners and providers (as well as policy makers) have a responsibility to take this on. Perhaps as resources become even more constrained, they might do so, but one can only hope they respond sooner.
There are great examples that could be built upon, across geographies and across industries. Credit card operators and banks handle massive data securely and could probably help understanding of the issues, if we asked. Likewise, the Rockefeller Foundation have already started to build on the remarkable benefits shown by universal health data systems in Abu Dhabi (which have achieved significant clinical and economic benefits - all published) in countries as diverse as Ghana and Indonesia.
The WHO, together with the ITU, have a programme to promote the application of information technology and are seeking partners. The International Diabetes Federation are joining forces with the GSMA (the global mobile phone trade association) to encourage diffusion at scale and are holding joint symposia at their congresses in Melbourne in December and Barcelona in February.
It would be great to see those responsible in this country easing the way for progress and for the commissioners and providers to grasp the opportunity before we find ourselves in even greater trouble.
David Oliver - you're asking a much bigger question which is why do public sector IT projects so often fail. I think the root cause is that the people running them from the public side are neither motivated by success or failure - civil servants are never sacked because a project fails. As Connecting for Health aptly demonstrates.
John Grumitt - your right to highlight that the problem is both solvable and being solved. My own company is building a solution delivering large scale, personalised diabetes care, based on the data, using exactly the same techniques we use to target and interact with customers for banks and retailers.
Comments
For most organisations their systems will be proprietary and therefore expensive to develop interfaces for.
The other complication is the range of organisations required - social care, primary care, mental health care, community care and secondary care all interact with each other, and in many cases multiple providers. The overlapping boundaries of the catchment areas (especially in London) make it very difficult to identify priorities for such integration, especially when the NHS organisations involved are subject to repeating reorganisations. This facet led to the development of a summary care record, only for this to be subject to concerns around data protection when operating on a very large scale. And in any event, the summary care record often does not provide the sort of detail that a team responsible for the shared care of a patient need to provide continuity of care.
Integration is especially challenging for such circumstances, where data entered once must be available to all - but only those who are directly involved in the current episode of care. Interfaces are generally easier to organisation on a push basis rather than pull - it is easier to download data from a system than to accept it from another. Reasons include 1) the difficulty of matching data - you may know me as Howell Huws, but that system has me down as Hywel Huws - even with the wonderful NHS Number and 2) the difficulty of matching data structures - my system may not have anywhere to put your specialised dataset. And then you get into how do you provide assurance of data integrity when data items may be changed by people on different systems simultaneously - whose is the master record? Hence the complexity and hence the expense of interfacing between the systems, and why it is so often put into the too difficult pile, and why so few actually get started.
An individual transaction-based approach (referrals, results, discharge notification) is likely to be easier to match to data protection issues, but does not resolve the shared input issue. Perhaps what is needed is a shared case record that only exists for the lifetime of the episode of care, contributed to by all organisations from data entered into their individual system and is then deleted when the GP decides (with the patient's consent, hopefully), that the episode is complete?
The problem is that organisations have chosen the wrong kind of solution. If you buy a cheap "black box", that ticked a form for "risk stratification", you can't then complain that the model is not open or adaptable. But then trying to build the software from scratch in open source is not the answer either - most NHS Trusts just don't have the resource.
The problems of building personalised, localised models of risk, and bringing together complex sets of data to build these models as identified by the commentor above, have been solved in other countries and in other industries. From Tesco Clubcard, to Kaiser Permanente, this is an understood problem with standard tools and approaches to solving it. The real challenge is that the NHS has yet to really understand how key this element is to ultimate success - although hopefully this blog is the start of that understanding. Buying a risk strat tool off the shelf is not the answer. And the start of that answer isn't as hard as some people think.
Second, I know that the current DH has made better informatics and a paperless NHS and person held transportable records a priority and key to the integration agenda and of course I agree with this. But is it a triumph of hope over experience when the last NHS IT project cost £8bn (the entire primary care prescribing or criminal justice budget) for what? How can we ensure its different this time. A triumph of hope over experience, driven by the fact that generally conservative ministers are more savvy when dealing with business/industry?
Just asking?
David
Healthcare has not yet embraced this opportunity. Perhaps the fear of getting it wrong, or the lack of experience of organisational change or indeed the desire to preserve the status quo in the face of limited resources and increased demand have all had a part to play.
Meanwhile, as a patient I want to see providers use my data to deliver services more efficiently and effectively yet am astounded at how slow they are to grasp the opportunity.
Commissioners and providers (as well as policy makers) have a responsibility to take this on. Perhaps as resources become even more constrained, they might do so, but one can only hope they respond sooner.
There are great examples that could be built upon, across geographies and across industries. Credit card operators and banks handle massive data securely and could probably help understanding of the issues, if we asked. Likewise, the Rockefeller Foundation have already started to build on the remarkable benefits shown by universal health data systems in Abu Dhabi (which have achieved significant clinical and economic benefits - all published) in countries as diverse as Ghana and Indonesia.
The WHO, together with the ITU, have a programme to promote the application of information technology and are seeking partners. The International Diabetes Federation are joining forces with the GSMA (the global mobile phone trade association) to encourage diffusion at scale and are holding joint symposia at their congresses in Melbourne in December and Barcelona in February.
It would be great to see those responsible in this country easing the way for progress and for the commissioners and providers to grasp the opportunity before we find ourselves in even greater trouble.
John Grumitt - your right to highlight that the problem is both solvable and being solved. My own company is building a solution delivering large scale, personalised diabetes care, based on the data, using exactly the same techniques we use to target and interact with customers for banks and retailers.
Add your comment