European data assets catalogue & eHealth Infrastructure - analyses of progress

We'll be looking at access to European patient-level data for research purposes over the next few months and have hit on a couple of online resources we thought worth sharing with you.

The first is a listing of data assets by country (and some international assets) with clear concise descriptions of their content.This is a great starting place for anyone looking to understand the large-scale datasets which might support various kinds of clinical, epidemiological and commercial research. The below is an excerpt, the entry for the Netherlands, to whet your appetite:

Netherlands

Health records

• Integrated Primary Care Information (IPCI) information from electronic patient records of 150 GPs covering more than 1 000 000 patients
• Pharmo Independent research organization for drug use and outcomes (including cardiovascular, metabolic disease, oncology and autoimmune disease, respiratory disease, and mother and child health). Overall it covers 2 million residents in the Netherlands and around 200 000 patients linked to GP patient records. Other data includes:
  • Community Pharmacy database (CPD)
  • Clinical Laboratory File (CLF)
  • General Practitioner database (GPD)
  • Dutch Pathology Registers (PALGA)
  • Hospital Pharmacy database (HPD)
  • Dutch mortality statistics (CBG)
  • National Dutch Hospital Registration (LMR)
  • Perinatal Registry (PRN)
  • Eindhoven Cancer registration (IKZ)

Health statistics

• GIP database from Health Care Insurance Board (Free online) covering outpatient drug utlization for 85% of population

The second resource is a series of reports by country, together with a summary across the continent, of eHealth Infrastructure initiatives in Europe. The final report is dated January 2011 but the speed with which these state and EU funded initiatives are able to progress is such that we suspect these analyses still stand and can be considered fairly current. The summary report is a good document to start with to understand the scope and purpose of the analysis but the individual country briefs are invaluable in their discussion of, for example, the data integration issues thrown up by Spain's regional administration, or the legal impediments to cross-border data sharing.

Clinical data - toward a single dataset supporting Research, Service Delivery and Performance Management

We were lucky enough last month to discuss with the Finance Director of a major London healthcare service, the use of enterprise data, from clinical to financial, to support performance management - and were struck by the extent to which the processes which support performance management analytics mirror those which support clinical research.

We were put in mind of our conversations earlier this year with Oracle and other vendors whose healthcare data warehouse platforms and attendant applications were being demonstrated as solutions to both research and performance measurement problems - this in turn called to mind the comments we had heard from data experts coming into healthcare from other industries who could not understand why information of all kinds pertinent to the administration of healthcare, from genomic analyses to staff costs, were not seen as belonging to a single and vitally important information asset.

The concerns of this London hospital in getting a handle on their data were strikingly similar to those of the research institutes we have spoken with - and indeed what they describe as performance management is really research by another name - when they correlate outcomes with treatment modalities and different packages of care, they are using their Business Intelligence architecture in many ways like a clinical research engine but with the addition of financial data .

Their core issue is data quality - they currently have four main clinical IT systems - one for each borough subsumed into their organisation. Trying to use these to derive information even at the level of 'number of patient encounters' has not been straightforward. Although they didn't go into it, we should imagine that supporting those clinical systems must be an array of systems capturing e.g. pathology, radiology, cytology data at a more granular level.

In addition, their financial data resides in three ledgers each with different coding - a 'consolidation nightmare' was how they described the move to a single ledger; painful but essential as they attempt to get a grip on expenditure.

The introduction of Service Line Reporting of income and expenditure (in order to assess profit by service) is driving their data validation and data quality improvement - but their clinical operations requirement to deliver an integrated service, developing 'packages of care' rather than looking at individual activities related to the same condition in isolation, is also dependent on quality and timely data - both financial and clinical performance management require the facility to benchmark accurately.

Touching on other IT issues their organisation faces, they mentioned that the mobile workforce are not well supported by technology - "it's still a surprise that no-one has developed a good mobile working solution for healthcare in the UK". Their words, not ours! They did suggest complications which we hadn't thought of hitherto, coming from a tertiary care background as we do, for example nurses on home visits may not be able to work online, thus need data stored locally to upload later which means on-device storage of personal identifiable data. We can't believe that is still an issue from a technological perspective, however, we can imagine that risk-aversion in respect of personal data in the healthcare industry is dampening demand for solutions - anyone care to offer a more informed opinion?All comments welcome!

We were really interested to hear how they addressed the development of their Business Intelligence capabilities - developing KPIs to meet significant and varied requirements from different commissioners. Their previous dashboards had been provided by NHS London - their goal at that time had been to ensure compliance with standards and regulations but as they have matured they now have to develop their own dashboards to meet more complex internally-driven reporting requirements. To do this, they have their own Performance and Information team who work on data collation and aggregation - creating Performance Packs - which provide detail by Service line under headings such as Operations, Quality, Finance, Workforce etc. supported by detailed analysis across the board from hard to soft data.

The capacity to present a performance summary across directorates has led to internal competition which is already leading to performance improvements. Their next steps?

the P&I teams are looking to automate the production of their performance packs and to create an overall dashboard for the organisation, leading in turn to a Balanced Scorecard.

Having realised that their previous KPIs and the systems which provided the data were inadequate, they embarked on a redesign of their processes by, and in this order!:

• Defining the goals / purpose / vision of the organisation


• Asking what information they need to support the delivery of these


• Asking what KPIs would adequately measure their delivery


• Then developing the systems which support the provision of the answers to the above

They are no longer looking simply at meeting regulatory reporting requirements - but at using their data internally to drive their performance - they are setting up data quality fora - having an external data quality audit and linking their output data to their income - and beginning to realise the benefits of placing data at the heart of the organisation.

Wales - what's their secret? Delivering successful healthcare informatics - care records

We blogged in the past about the fact that Australia, Scotland and Wales had stolen a march on the English in developing data linkage facilities for healthcare research and had heard this blamed on the heterogeneity of our healthcare systems. We now read in eHealth Insider that "Wales is also having success with sharing patient data via its Individual Health Record. More than 300 GP practices have switched on access to the record, which is available to emergency care providers. It includes demographic information, medication, allergies, test and x-ray results, and medical problems from the past two years."


A great comment on the eHealth Insider article compares the modest investment behind the Welsh achievement to that behind the English National Programme for IT with its "modest return".

"It's hard to comprehend that the relatively small sums discussed here have resulted in the extent of progress made so far in Wales; particularly when we consider the largesse associated with the English NPfIT [for] such a modest return."

Patients' views on participating in medical research - Part 2: Engagement and Participation

Our last post here began our summary of recent work on patients' attitudes to participating in medical research - part of our primer on the current state of play in clinical research in the UK selecting choice elements from recent reports; check out the footnotes for interesting sources to follow up. We include in this a look at the influence of media coverage of issues pertaining to data security on patients' attitudes and behaviour.

As always, let us know what you think - if there are more recent / complete / credible studies out there which draw different conclusions, let us know!

Engagement and participation

The UK has a long history of public support for health research, as evidenced by the large number of participants in clinical trials and population studies (For example, the UK Collaborative Trial of Ovarian Cancer Screening and UK Biobank have recruited their targets of 200,000 and 500,000 individuals (respectively) with minimal objection to the use of their healthcare data) and the generous contributions to medical research charities such as Cancer Research UK and the British Heart Foundation.[1]

Public engagement initiatives in relation to specific issues, such as the use of patient data, generally show that research is warmly supported. The attitudes of over 1,000 adults towards participating in health research were examined in the Wellcome Trust Monitor survey. Seventy-one per cent of participants indicated that they would be willing to give blood or tissue samples for research and 62% were willing to test a new treatment for a disease from which they were suffering.[2]

Evidence from two national research studies demonstrates that a small number of patients complain about receiving direct invitations to participate in research. The UK Collaborative Trial of Ovarian Screening is one of the largest ever randomised controlled trials, covering 13 NHS Trusts in England, Wales and Northern Ireland, with successful recruitment of more than 200,000 women. Of the 1.2 million women invited to participate in the study only 32 complained about being contacted. UK Biobank reported from its integrated pilot phase that approximately 1 person from 1,000 invitations indicated that they did not want to participate because of concerns that their contact details had been provided to UK Biobank by the NHS.[3]

There are a large number of organisations working to improve patient and public engagement with health research, including (but not limited to) UK Clinical Research Collaboration (UKCRC), INVOLVE, regulators themselves, the medical Royal Colleges, research charities and disease specific patient groups working to help the public understand the role and importance of research as an integral part of the care system. 

Media view – data protection

The influential role of mass media has important implications for the formation of public opinion and consequently public behaviour and the actions of policy makers. The most significant impact on attitudes towards the storage, transmission and use of personal data in healthcare is made by coverage of breaches of regulations and guidelines.

Though many of the stories do not relate to data used in research per se, their impact contributes to patients’ concerns about any use of personal medical data.

Storage and transmission of data are key to research, many large datasets (for example the national disease registries) inducting data from a variety of sources and releasing data for research to geographically dispersed users. A key aspect of the conduct of research is the ease with which those researchers can receive the data. The choice of transmission method is not driven solely by actual risk analysis: while an encrypted DVD has a high level of innate security, public perception of sensitive information being moved around on DVDs, memory sticks and laptops is an important consideration. In fact it was a major issue identified in the UK Ministry of Justice's report on Data Sharing, 2008.[4]

A recent article from eWeekeurope.co.uk backs up perception with data under the inflammatory headline “A Freedom of Information request by… Software AG has revealed that most public sector bodies have no idea about secure data transfer.”[5] The article cites recent examples of the loss of sensitive information by public bodies: “A couple of years ago, Her Majesty’s Revenue and Customs (HMRC) lost a number of CDs containing private information on thousands of people. But there have been many more recent examples. Last July the UK Ministry of Defence admitted it had lost an entire server from a secure building – as well as 1.7 million individuals’ personal data. In November the UK Rural Payments Agency (RPA) lost backup tapes containing the payment and banking details of 100,000 farmers in the United Kingdom. And only last month an NHS worker in the secure mental health unit of a Scottish hospital was suspended, after he lost a USB stick containing patients’ medical records. The USB stick apprently contained unencrypted sensitive information – including the criminal histories of some violent patients at the Tryst Park unit at Bellsdyke psychiatric hospital. The stick was later found by a 12-year-old boy in the car park of an Asda supermarket.”

The NHS was recently (April 2010) revealed by the Information Commissioner’s Office (ICO) to be responsible for the highest number of serious data breaches of any UK organisation since the end of 2007. David Smith, deputy commissioner at the ICO told the Infosec security conference the NHS had highlighted 287 breaches to it in the period, accounting for more than 30% of the total number reported.[6] Most of the breaches were the result of stolen data or hardware, followed by 82 cases of lost data or hardware. Richard Vautrey, the deputy chair of the British Medical Association's GPs committee thinks the number of breaches reflect the size and complexity of the NHS (the UK's largest employer with 1.7m staff) as well as its culture of openness.[7] Whilst comments in the BBC’s coverage mention in mitigation that the public sector’s culture of reporting all breaches contrasted with the private sector’s behaviour, these do little to lessen the impact of the headline: “NHS worst for data breaches.”

---

[1] The Academy of Medical Sciences: A new pathway for the regulation and governance of health research

[2] Ibid.

[3] Ibid.

[4] Ministry of Justice: Data Sharing Review, 2008 [Richard Thomas, Information Commissioner; Dr Mark Walport]

[5] http://www.eweekeurope.co.uk/news/public-sector-clueless-on-secure-data-transfer-7635

[6] http://www.bbc.co.uk/news/10089066

[7] Ibid.

Data Integration platforms - Research and Service Delivery - Oracle

We're back to full strength now and looking forward to passing more in-depth info your way this year than hitherto - starting with a look at data integration platforms (Oracle, Orion, IBM, Cerner, OpenClinica and more) being considered for use in UK and US healthcare institutions looking to harness the power of their currently fragmented data to deliver improved services and support research.

We recently started working with a major London-based cancer research centre who are entertaining pitches from various suppliers looking to pull together their pathology, cytogenetics, radiology and clinical data (plus everything else, immunology, toxicology, virology, you name it) and we've been digging into the detail behind the glossy slides - starting with Oracle:



Oracle have an increasing presence in this space in the US and are looking to use their learning there to expand their healthcare division in the UK. Their current UK consulting workforce (healthcare) is still in the single digits, but they sought to impress with details of their R&D spend and some recent examples of their work in the States. For some interesting discussion around their presentation of R&D spend, see these two bloggers (enterpriseirregulars and martijnlinssen) who discuss absolute spend vs. proportional spend and what it means about Oracle's R&D budget compared to, for example SAP.

Key to their presentation was a slide describing the systems architecture of a completed end-to-end solution with the Oracle Health Data Warehouse Foundation at the core and an 'Omics analysis platform linked in which we thought was pitched as the value-adding component.

Certainly the 'Omics platform is indicative of their aim to get to work in the translation medicine area (described as the North Star of their current thrust) - but it's an undeveloped product, not even fully implemented at Moffitt as far as we are aware. And that will be its first implementation - Moffitt hinted that this was the missing component in the Oracle solution the first time round (the guys from Oracle, however, countered that at least their core integration engine worked where Microsoft and Orion had both previously faltered).

See here for an interesting article which describes how a number of other vendors may be brought in to perform analytics on the core Oracle database.

As it happens, Oracle are also developing a Translation Research Center product - see here for Oracle's full healthcare product offering.

We're familiar with some of the Oracle products already - the Health Transaction Base, it's Enterprise Terminology Services - the latter is in use at at King's Health in a their enterprise oncology solution being implemented by IDBS (see our blog entry) - but King's Health are using ORION's Rhapsody data integration product rather than an Oracle solution - we will try to get answers from Peter Parker as to their rationale.

We had a few issues with their presentation which centred on that solution architecture slide which featured the Oracle solutions centre-stage in bright red, and off to the left in grey was the data extraction process from the multiple systems currently in place in any healthcare provision organisation. They admitted that this small grey box is where the majority of the work and cost is incurred!

Of note, however, they are developing a series of app-style products which can then plug into the clinical and genomic data warehouses to fulfil various research and service delivery ends - looks interesting and we will try to get you more info on this.

As imagined, it looks like the major pain point in any implementation is extraction from clinical systems such as those Cerner and others are currently managing in the UK - we're going to be in touch with some folks who have developed bespoke SQL products to perform exactly that task in the coming weeks - keep checking here for more on this - and we'll take a look at some of the other products in this 'space' - so watch this [space].

For a discussion of the Dana Farber's experience implementing Oracle - see here.

Proposals for a Research Support Unit

Apologies for the lack of posting in the last week - we've been working on two new initiatives, supporting a leading UK cancer research institute with vendor selection for research and service delivery systems - and working on a proposal for a Research Support Unit for UK based healthcare institutions providing exactly that service nation-wide: support in the process of selecting and preparing for solutions integrating their data for both operational and research ends.

If you are UK-based and require support, particularly in collating clinical data from multiple hospital systems (pathology, radiology, cytogenetics, PAS) in order to support research or service delivery - leave a comment with your details or email us at bioinformaticsolutions @ gmail.com and we'll discuss how this proposal could be used to help your institution.

How Computer Science Agents Can Help in Medical Informatics

We've been trying to sneak them under your radar, almost certainly without success, but here in broad daylight we declare: the last of our updates from the NOCRI Information Systems workshop - and we have mercifully little to say on the talks as Prof. Michael Luck's abstract says it all [Prof. Luck is the Head of the Department of Informatics at King's College London]. Our only comment is that it might be worth checking out the PDF "50 facts about agent-based computing", which Prof. Luck mentioned, to consider some current applications of computer agents across a variety of industries. His abstract is presented here:

“Agents are computer systems capable of flexible autonomous action in dynamic, unpredictable, typically multi-agent domains. In particular, in dynamic and open environments, heterogeneous systems must interact, span organisational boundaries, and operate effectively within rapidly changing circumstances and with dramatically increasing quantities of available information. Thus, the need for some degree of autonomy, to enable components to respond dynamically to changing circumstances while trying to achieve over-arching objectives. For example, in the domain of bioinformatics, determining protein function can involve multiple independent processes connected together, operating on multiple independent data sources, in some workflow. In this context, multi-agent systems, in which the various processes and data sources are encapsulated as agents that interact with each other, can provide a means to manage primary databases, perform sequence analyses (using existing tools), or store and present resulting information in a coherent fashion. In the domain of e-medical records, multi-agent systems may similarly provide the means to integrate disparate processes and data sources. Rather than offer new methods for performing these tasks, this instead organises existing methods for effective and flexible operation.”

That wraps up the NOCRI workshop except to say that the two concrete Next Steps to come out of it sound promising: the first being the organisation of a co-ordinating group to further knowledge sharing in informatics and the second being the creation of a web-resource for learning materials.

NIHR / NOCRI Information Systems Workshop - Part 2 - FP7 TRANSFoRm: Translational Medicine and Patient Safety in Europe

Next up, after Mike Denis, in the first session was Prof. Brendan Delaney (Guy's and St Thomas' Charity Chair in Primary Care Research, King's College London) talking about the TRANSFoRM project under FP7 – for the uninitiated, “'Framework programmes' (FPs) have been the main financial tools through which the European Union supports research and development activities covering almost all scientific disciplines.”

The project is looking to develop infrastructure supporting research and diagnostic decision support in primary care – and in the UK is working with GPRD (the primary care data-source). 

Prof. Delaney referred to the well worth reading report from the US Institute of Medicine of the National Academies DigitalInfrastructure for the Learning Health System: The Foundation for ContinuousImprovement in Health and Health Care - Workshop Series Summary 

 

Prof. Delaney’s big statement was that “Research is in crisis!”, his verdict on the recent reports (we will find the link!) on caBIG was that the lesson to be taken is, “Don’t build massive infrastructure projects! They just grow and grow and create wealth for consultants.” Interestingly, other speakers noted that they were using at least one caBIG tool in the development of their own infrastructures.

Prof. Delaney’s work on TRANSForM encompasses a data quality tool, data linkage functionality (browsing, selection, extraction) and semantic mediation – the latter facilitating the transformation of Electronic Health Records (EHRs) into research quality data (EHRs with tabs for CRFs which partially pre-populate and can ultimately be signed-off online).

This last facility is at the heart of the work, using reference ontologies, controlled vocabularies etc. to interpret the EHRs.

The first year’s deliverables (including the privacy framework, security framework and provenance framework – together creating the governance framework) are already available on the website: http://137.73.82.45/Home.html

We’d be interested to know the relative ease of navigation of the regulatory landscape for each of the involved nations.... and will the deliverables form the semantic mediation be published?

This is the abstract for Prof. Delaney’s talk:

“TRANSFoRm is an EU FP7 project that aims to develop and test a digital infrastructure for the 'learning healthcare system'. Specifically we are working to develop capacity within primary care electronic health records to manage research workflow (recruitment and follow up), data collection and record linkage, and to develop diagnostic decision support using a common infrastructure.

NIHR / NOCRI Information Systems Workshop - Part 1

With apologies for the delay in getting this to you, here comes the first of our updates on last Monday's NIHR Office for Clinical Research Infrastructure (NOCRI) Information Systems Workshop!

The workshop was held at Guy's Hospital and the introductory remarks began with a welcome from Professor Simon Lovestone, who is, in addition to being the Director of the Mental Health Biomedical Research Centre at South London and Maudsley & King's College London, Director of Research at King's Health Partners Academic Health Sciences Centre.He was followed by Mark Samuels (MD, NOCRI) and Peter Knight who is now Deputy Director of R&D and Head of Research Information and Intelligence at the Department of Health.

Session 1 - Chair: Prof. Peter Parker, Cancer Theme Lead, NIHR BRC at Guy's and St. Thomas' NHS Foundation Trust & King's College London.

We've blogged twice before about Mike Denis's presentation covering South London and Maudsley NHS Foundation Trust's (SLaM) Clinical Records Interactive Search (CRIS) system - Mike was up first in the first session after the introductory messages and talked on the Vision for Translational Informatics at SLaM.

He talked to two themes in relation to SLaM's creation of a Personal Health Record, the first being the technical aspects (including deep pseudonymisation - meaning that it extends into the free-text within records) and the second being the social aspect that puts the patient in the driving seat of the process but also aims to link Patient Reported Outcomes to research.

An interesting aspect of the technical discussion was the use of intelligent agents to provide automated analysis and predictive analysis - which chimed with Prof. Michael Luck's comments later in the day on the use of computer sicence agents in medical informatics - of which more later.

Mike emphasised the scalability of the approach, the potential to implement it across multiple centres and carry out data linkage inter-organisationally.

Here's the abstract provided for Mike's talk:

“Within SLAM, we have sought to focus on a relatively fine-grain model of clinical informatics, building the Clinical Records Interactive Search (CRIS) system which provides researchers access to full but anonymised data on a large mental health NHS provider covering a 1.2m population catchment area and containing over 180,000 case records. Key areas for resource development which will be outlined include the routine collection of consent for researcher contact, informative data linkage with internal (biological) and external (health and social) data resources, and adding depth to the dataset through natural language processing, shared health records and patient reported outcomes. We believe that these developments, coupled with advances in nested trial designs and electronic health records (EHR) feedback loops will help move the current focus of EHR and 'case register' research in our field from an observational to translational footing.”