We have defined [1] a methodology for extracting patient centric data from multiple heterogeneous hospital information systems (IS).  The collection of data on patients with prostate cancer has been one successful example of our approach and has establised proof of principle for this proposal.   The study involved collecting data from (i) textural reports from the Picture Archiving Communication System (PACS); (ii) clinical coding and appointment information from the hospital Patient Administration System (PAS); (iii) details of diagnosis, treatments and history from the oncology department system (ONC) and (iv) PSA test results and histopatological reports from the biochemistry and histopathology system (LAB).  Additionally, data from the cancer registry served as a quality cross-check. Such data integration is not currently the norm unless patients are part of a clinical trial. Prostate Cancer has been estimated to directly cost the NHS £45 million per year in 1997 [2]. It is the most common male cancer in the UK and there is a need for data integration because there is considerable debate about which treatment is best.

After collecting the data and generating an Operational Data Store (ODS) containing patient centric information on thousands of patients with prostate cancer, we have proposed a framework for the construction of authoritative e-pathways from such data.  We have developed measures of quality of each pathway constructed so that clinical analysis can include only those records deemed of sufficient quality.  We have also developed software to build, visualise and explore the e-pathways.  This work, now being published, will enable further clinical and epidemiological research on prostate cancer but also the development of methods to compare and analyse pathways, improve integration and visualisation enabling clinicians to follow their patients' journeys.  The database assembled is unique in the content and presentation of the information on prostate cancer pathways. 

A number of challenges still remaining.  First, retrieving, linking and collating patient centric data is non-trivial.  We need to implement a system that will periodically collect, link and analyse new data and add it to the ODS.  Second, although we can now present e-pathways for over 1,900 patients, we would like to build capability to conduct analysis such as compliance with the NICE defined prostate cancer pathway, survival of patients on different treatments or ability to match a patient's e-pathway with other similar e-pathways to look for possible outcomes.Finally, we believe that it is possible to extend all our work on building and analysing e-pathways to other diseases.  Another case study to assess the generalisability of the tools developed  is that of patients with stroke.  Stroke is  particularly suitable because the timing of events in the disease pathway is of very importance and there is great interest in constructing  and analysing such data in the local NHS hospital.  Stroke represents a significant economic cost to healthcare providers and society as a whole. In 2006, the direct, and indirect cost and cost of informal care of stroke in the UK to the National Health Service (NHS) was estimated to be £8.3 billion [3].

Bettencourt-Silva J, de la Iglesia B, Donell S, Rayward-Smith V. On creating a patient-centric database from multiple Hospital Information Systems. Methods Inf Med. 2012;51(3):210-20.

Chamberlain J, Melia J, Moss S, Brown J. The diagnosis, management, treatment and costs of prostate cancer in England and Wales. Health Technol Assess. 1997;1(3):i-vi, 1-53.

Allender S, S.P., Peto V, Rayner M, Leal J, Luengo-Fernandez R and Gray A, European cardiovascular disease statistics. European Heart Network: Brussels, 2008

Huang Z, Lu X, Duan H. On mining clinical pathway patterns from medical behaviors. Artif Intell Med. 2012 Jul 16. [Epub ahead of print] PubMed PMID