Following on from our recent article, Why Real-World Evidence? (which can be found here). This week we take a look at the what data is available for medical research. We hope this will be a useful framework to utilise when trying to decide how best to answer your questions. The next obvious step is to try and work out how to answer them. The simplest way of doing this is to work out what variables you need to test your question, decide on a rough budget, and then consider the options. From my point of view, I find the most effective prism for this assessment comes as the Four Domains of data (see below) which includes all data types and sources possible.
Before Start of Study
Retrospective data includes all data ever recorded anywhere on a given patient. Using a Randomised Controlled Trial (RCT) to define the term (in blue), this would normally be the patient’s medical history. However there is no reason why this can’t be the entirety of the patients integrated care record (primary, secondary and tertiary care). In fact any data held electronically on the patient, if it can be linked back to the patient, is theoretically in play (as illustrated in green).
Patient Interaction data is any information actively collected for a study on a patient. Again, using the RCT to define the term, these are your clinic visits where you take measurements of the patient whether that be physical and patient reported. Of course the scope of this data is the same, whether it is an RCT or RWE.
Between the study visits
Patient Tracking data is any information collected passively on the patient during the period of interest. In an RCT this would usually be limited to Adverse Event reporting, however there is no reason why this should not be extended to the entirety of the patients medical record for the duration of the period of interest, every interaction with a healthcare professional and every prescription taken out. Coupled with this, the whole realm of wearables and social media monitoring can be utilised here as well.
After the end of the study
Outcome data is any information collected on the patient after the end of the period of interest. This is the domain usually missed from an RCT. However, there is no reason why the patient should disappear off into the ether at the end of the study. Simply by consenting and tagging the patient’s care record during the study it is possible to collect all the data outlined in the previous three domains for follow up work after the study.
It is from these Four Domains of Data that all study methodologies are possible. Examples of the some styles can be seen below:
To bring this into a more practical realm, the key considerations in relation to budget are; ease of access to the data and, whether patients have to be consented (highlighted above), or whether the data can be given on behalf of the patient by the doctor (in an anonymised form).
A common missed opportunity is only collecting the data required for your current question, rather than taking advantage of the opportunity to build a broader data set. In the pharmaceutical world it is nearly always inevitable that further research will be required into a given treatment or condition (even if it is just a Phase IV, post authorisation study to help promote sales). By applying a little more budget early on subsequent studies can be radically reduced in price because the data will already have been collected and sitting on a server. A simple example being to collect the patient EHR at the beginning of the study to supplement their patient history. Once this data is added to the study data set, it is then available for secondary analysis as many times as required, potentially for completely unrelated topics.
Final consideration is around the quality of the data. It seems like an obvious point, but not all data is created equal, so to speak. All sources have some level of bias and inaccuracy inherent within them which should be considered. A useful rule of thumb can be seen in the diagram below from Thy Do of Takeda (see ‘Trends’ issues ‘Real-World Evidence Revolution’).
Within the above, are the building blocks required to answer any question you can possibly ask about any treatment or condition.
For more information about our research services, email us at firstname.lastname@example.org or call on +44 (0) 1438 768718.