Can electronic health records improve the efficiency of large clinical

Can electronic health records improve the efficiency of large clinical trials?
Supervisors:
Associate Prof M. Angelyn Bethel 1
Associate Prof Louise Bowman 2
Prof Simon Griffin 3
Departments:
1
Diabetes Trials Unit, OCDEM, Radcliffe Department of Medicine
Clinical Trial Service Unit, Nuffield Department of Population Health
3
University of Cambridge Institute of Public Health
2
Background
In diabetes, new antidiabetic agents must demonstrate both glucose lowering and cardiovascular
safety before approval. Clinical outcomes mega-trials are required to provide robust assessment of
the risks and benefits of treatment. A typical mega-trial is initiated outside routine health care in
selected cohorts and requires an extensive and expensive trial machinery to provide multicentre
scientific, ethical, and regulatory oversight. To limit costs, the trial are large but with short durations
of follow-up, measure surrogate or limited outcomes, and often leave important long-term riskbenefit balance questions unanswered. For change to be possible, more efficient trial methodologies
are necessary.
Electronic health records (EHR) and routine data sources are proposed tools to increase efficiency
and reduce trial cost. EHR advantages include providing long-term multidimensional follow-up of
disparate outcomes (e.g. cancer, fracture) and, when process data can be obtained (e.g vital signs,
laboratory values), obviating the need for face to face visits. However, the datasets are complex and
may not reliably ascertain events due to inconsistencies in diagnostic coding and treatment practices
across a health care system. Use of EHR to replace established patient follow-up practices must be
validated to ensure scientific integrity.
Hypothesis
EHR can be used to reliably identify important 1) cardiovascular outcomes, 2) diabetes-related
outcomes and 3) safety outcomes, thus allowing cost-effective methods of follow-up in long-term
outcomes trials in diabetes.
Aims
Using data from completed large scale outcomes trials, including but not limited to the United
Kingdom Prospective Diabetes Study (UKPDS), the UKPDS post trial monitoring study (UKPDS-PTM),
and the Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes
in Primary Care in the UK (ADDITION-UK), event ascertainment from traditional clinical trial practice
will be compared with that achievable from EHR for cardiovascular events, microvascular
complications of diabetes, and other safety events. The design of these trials will allow comparison
of EHR ascertainment with face to face visit and/or questionnaire ascertainment protocols.
Description of Work
1.
Obtain EHR dataset and prepare it for research use. The fellow will participate in a workshop to
facilitate the significant upfront learning required to understand and use EHR data. Issues
addressed are the complex coding of data items, data provided at a level which is not
immediately amenable to analysis, missing data, duplicates, and costing episodes via Healthcare
Resource Groups (HRGs). Workshop learnings will be applied to data across multiple clinical
trials to prepare datasets for research use.
2. Evaluate EHR for event ascertainment. Identified EHR outcomes of interest will be compared to
data collected from completed clinical trials. Validation in multiple datasets will allow
comparisons of EHR to clinic visits and/or questionnaire follow up protocols. Events of interest
are cardiovascular, diabetes, and safety outcomes. Benchmarks for comparing EHR to traditional
trial methods are:
 Ascertainment. The number of events of each type identified.
 Concordance. The agreement between investigator-reported event types and EHR. We will
also examine concordance with adjudicated events, where available
 Cost. We will compare cost of the EHR event ascertainment approach (including approvals
and application fees, data processing and analysis) to standard trial practice (including site
payments, monitoring requirement, and adjudication activities).
3. Evaluate the performance of EHR in real time. In event-driven trials, the time taken for event
ascertainment affects trial duration. If EHR follow up is shown to robustly ascertain important
patient outcomes, these methods will be applied in an ongoing outcomes trial to ensure that
event data can be collected in a timely manner. We will compare time lags between event
occurrence and traditional investigator reporting dates to those between event occurrence and
detection in the EHR to determine if delays or efficiencies are introduced. As above, the cost of
using EHR for real time event identification, which will require multiple data downloads (e.g.
monthly), will be evaluated.
Recent relevant publications
1. Holman RR, Paul SK, et al. (2008) 10-Year Follow-up of Intensive Glucose Control in Type 2
Diabetes. N Eng J Med 359: 1577-89.
2. Griffin SJ, Borch-Johnsen K, et al. (2011) Effect of early intensive multifactorial therapy on 5-year
cardiovascular outcomes in individuals with type 2 diabetes detected by screening (ADDITIONEurope): a cluster-randomised trial. Lancet. 378: 156–67.
3. Bulbulia R, Bowman L, et al. (2011) Effects on 11 year mortality and morbidity of lowering LDL
cholesterol with simvastatin for about 5 years in 20,536 high-risk individuals: a randomised
controlled trial. Lancet. 378(9808): 2013-2020.
4. Aung T, Cowan H, et al. (2011) Recruiting patients cost-effectively by mail. Trials. 12(Suppl 1):
A117.