Research Programme 1: EPR Support of Clinical Trials

Prof Ian Ford, Dr Allan Gaw, Prof John Norrie, Prof Frank Sullivan

Aim

  1. To evaluate the way in which EPRs can best support a range of clinical trials and the subsequent conduct of those trials.

Support of Clinical TrialsBackground

As the costs of phase 3 and 4 clinical trials spiral, EPRs provide the opportunity to enhance the feasibility and efficiency of drug and technology development. We will explore two key areas.

  1. Research into data quality: Clinical trials require good quality data and clinical endpoint adjudication is a prerequisite of good design. Others have questioned the added value of this labour-intensive and expensive activity. We have recently established proof-of-concept for using EPRs within trial and post-trial follow-up within the WOSCOPS trial for broad classes of coronary events and for all-cause mortality. Inconsistencies can arise however because of ascertainment and reporting error, a lack of specificity in coding as well as record linkage errors.

    Research questions:

    1. Can we extend the validation of routinely collected data for more general use as endpoints in major trials?
    2. Can we identify eligible patients for secondary prevention studies by extracting data from stroke unit databases, the SCI Store databases including all laboratory data and with ongoing follow-up from linkage to SHIS-R?

  2. Key components of this work package are to make use of in house datasets for approximately 14000 UK patients (approximately 10000 recruited in Scotland) from the WOSCOPS, IONA, PROSPER and IMAGES trials for whom we have detailed long-term clinical trial data and who can be record-linked to NHS datasets to enable comparison of linked events and events identified using traditional methods including the use of adjudication committees. This will allow comparison of the different approaches for the identification of specific outcomes of interest (in particular cancers/major cancer subtypes, stroke/stroke subtypes, congestive heart failure, myocardial infarction as distinct from unstable angina and other hospitalized angina, mismatches of cause of death associated with death due to infection)
  3. Using EPRs to identify potentially eligible participants for clinical trials and to follow-up within and beyond the period of the original study: We have recently built on pilot work conducted within the MRC funded e-Science project VOTES and have developed study specific software for the major clinical trial SCOT that searches GP electronic databases for eligible patients for the trial and facilitates generation of invitation letters. The software also extracts data on prescription medications from these databases throughout the trial.

    Research questions:

    1. Can we develop a generic application, to support naturalistic clinical trials in general practice within the NHS, to identify eligible patients from GP systems and extract and upload specified data on an ongoing basis, for those consented into a trial, focusing initially on the GPASS system and then moving to other proprietary systems?
    2. Can we develop applications in secondary care that will identify eligible patients for secondary prevention studies by extracting data, from stroke unit databases and the SCI Store databases containing all laboratory data and with ongoing follow-up from linkage to SMR data?

Methods

  1. We will develop an application for use in general practice that will identify potentially eligible study subjects in their electronic patient records by inputting researcher defined criteria consistent with a trial's inclusion/exclusion criteria. The application will also have the ability to produce trial invitation letters on practice letter heading and extract clinical data form consented subjects. Longer term data extraction of READ coded from GP records using Scottish Enhanced Functionality software will also be linked to SHIS-R.
  2. Develop an understanding of stroke unit databases and subsequently build an application that will identify potential clinical trial participants by linking stroke patient data with routinely collected secondary care datasets.

Deliverables

Publications will determine the completeness, accuracy and utility of routinely collected data derived from EPRs for identification of study subjects and end point assessment. Software applications will also be developed for researchers to identify potential study subjects in primary care and stroke units.