By Stuart Cook, director of CDM (clinical data management), Quanticate.
Do you remember where you were when you heard the slogan, “cleaner data faster?” That was the promise of electronic data capture (EDC). EDC is now widely embraced and is often the de facto method of data capture in clinical trials. How far can EDC expand? We already have robust processes and services in electronic patient recorded outcomes (ePROs), and this area has combined well with certain types of EDC studies. The question is, “How far are we from the utopian vision that is ‘e-Clinical’ where data is recorded once and is available directly in the database?” Who is responsible for reviewing the data in an eCRF (electronic case report form) against the source data? What if we didn’t have eCRF data — just source data? What if the data recorded in source data were transferred to the database in real time?
EDC studies (i.e. eCRFs) are arguably a contemporary version of a paper CRF with a paper source document. The assessment is recorded in the source at the site, entered into an online EDC application, and then verified by a monitor. It is rare that any development of an eCRF involves the ultimate end users, the investigator, and study nurse in the testing. Therefore, sites often are faced with entering data in a different way from how it was originally captured. Surely there would be greater buy-in from sites if they entered data directly into a repository that replicates the format of the current paper source or electronic medical record (EMR). If this could be accessed and utilized directly at the patient bedside, then data could be accessed in almost real time by the sponsor.
The rise in database variable naming and form structure standards has led to increases in data quality and greater adoption by study teams. As these electronic data standards move into other areas of data capture, we have seen the steady adoption of ePROs as their use negates late patient data entry and allows for reasonably continuous access to patient data via online tools.
We also have seen the integration of data that negates reconciliation. If data that is already captured in an IVRS (interactive voice response system) or IWRS (interactive web response system) is used to populate fields in an eCRF, entry errors and reconciliation can be reduced or eliminated. These processes also can provide a mechanism, through integration, that halts data entry until the patient is acknowledged as randomized within the system, thus negating erroneous entry of screen failures.
Integrating Source Records With EMRs
It’s often difficult to integrate electronic source records with EMRs because there are two distinct objectives in terms of the data captured. A trial captures data based on a protocol and in a system designed around a CRF. If clinical patient data was collected only as electronic source data, it would be possible to integrate that data with an EMR using healthcare interoperability standards such as Health Level Seven (HL7) data export or, alternatively, printing the electronic source data and adding it to the patient’s medical record.
The harnessing of all electronic data could move a trial toward being truly “e-Clinical.” However, there are still questions about the legitimacy of data as it is transferred from one location to another through human data entry. With source data verification (SDV) and data management review through edit checks and listings, transcription errors are found throughout trials. If there were no transcription of source data from patient records into the CRF, it wouldn’t be necessary to monitor for transcription errors. There would still be a requirement to manage transcription errors, but it would be at the point of initial entry (e.g. checks for future dates, ranges checks). This approach could be harnessed in certain types of trials where there is no other papersource collection, such as local lab records or paper ECGs, as these would be additional primary source data that would require entry.
Finally, we already have seen attempts to reduce the cost of monitoring visits by increasing central or remote monitoring and targeted SDV. Even so, SDV contributes around 50% of a monitoring visit. The current model of using paper documents or EMRs at the site to record patient source data means that in order for the data to be monitored, the monitor must be able to access the source documents. If electronic source records were used, there would be no need for reconciliation between the source data and the CRF data. Potentially, the frequency of monitoring visits could therefore be reduced.