Reuse of health care data for scientific research
Health care data may not be reused without respecting patient privacy and autonomy. All conditions for reuse of health care data for the purpose of scientific research are therefore described in the SOP Reuse of health care data for the purpose of research, which translates relevant legislation into a concrete procedure across Amsterdam UMC. The SOP and corresponding forms can be found under Procedures and templates.
Before data, an image, a report or a cohort identification can be released, the study must have been reviewed by the METC or a healthcare evaluation must have been ordered. Submit the request form for data extraction to the RDM helpdesk.
Below is a list of the available tools and the corresponding areas of application:
Structured health care data stored in Epic and other data sources can be made available for research through the Research Data Platform (RDP). This RDP consists of clusters of similar data, such as medication, care paths, labs, and more. These clusters are defined identically at both Amsterdam UMC sites, allowing for combined data reuse. Data are retrieved from Epic as well as non-Epic and pre-Epic sources.
Discrete/structured data can be searched, as can notes/reports such as radiology and pathology reports. The data from these files can then be imported and analyzed with various statistical packages such as SPSS, R and STATA.
For an overview of all available data in the RDP, see the RDP catalogue.
Data not yet included in the RDP could also potentially be reused for research. However, delivery of such data is not standardized and therefore more laborious. Epic forms let you capture research data in a structured manner, making them easier to extract later, which may be desirable in long-term registrations in which these data are part of the care process. Data can be extracted from these forms, which can also be imported into Castor EDC.
In consultation with a RDM Helpdesk employee the target population will be selected based on the desired clinical inclusion and exclusion criteria (e.g. whether the images were made before or after an intervention). The RDP then provides a cohort with patient numbers and image identification numbers (so-called accession numbers) for the image team of the EvA Service Centre or the Radiology and Nuclear Medicine department.
Activities of the image team
The images are anonymized or pseudonymized, using special software (Clinical Trial Processors, CTPs), before they are released for clinical research. Click on the following link for an overview of anonymization and pseudonymization profiles: Decision tree profiles for anonymization and pseudonymization.
Apart from the images, digital image data are stored in a database that can be queried. These data can also be provided at the request of the researcher.
On the one hand there are the digital images themselves and the corresponding data and metadata. On the other hand there are the clinical data of the population the researcher is looking for.
Various reports can be created within Epic, e.g. a list of current patients meeting a specific criterion or a report that can be generated with certain settings (age, gender, department) that link data and then present it in a PDF document. These reports can also be used to find potential study participants.
It is also possible to export the results from a Crystal report to a separately authorized file share, in the formats Excel and CSV.
A more detailed explanation and more examples can be found here (link follows).
To set up Epic reports, ask your department's Epic Power User or contact the RDM helpdesk.
To determine whether a study is feasible, it may be necessary to count or identify patients who are potentially eligible for participation. To avoid having to do this manually, which is both very time-consuming and suboptimal from a privacy perspective, Epic can be searched for this automatically.
There is a software application available that helps to gain insight into the size of specific patient populations within Amsterdam UMC, SlicerDicer. After specifying structured search criteria (e.g. diagnosis, medication, department), RDM can provide a list of patients meeting these criteria.
All researchers with access to Epic are able to use Slicer Dicer as a cohort identification tool. After specifying structured search criteria (e.g. diagnosis, medication, department) a list of patients is created in a given time frame. No patient numbers are available though. RDM can issue the data after all data request procedures are finalised.
Creating patient overviews
SlicerDicer is an Epic tool with which you create your own patient overviews. SlicerDicer is a powerful data exploration tool for clinical questions. Using SlicerDicer, you can easily search Epic for your intended study population. SlicerDicer also lets you explore trends by easily refining your search queries. Trends can also be explored at the level of individual patients.
For an overview, see the wiki pages on the website of the EvA Service Centre:
- Epic:SlicerDicer data models
- Epic:SlicerDicer and adding patients to a patient list
- Epic:SlicerDicer advanced tips
- Epic:SlicerDicer manual
- Epic:SlicerDicer access and permissions
Notifications for eligible new patients
Best Practice Advisories (BPAs) are real-time pop-ups that can alert clinicians to a potential study participant based on patient characteristics such as, medication use, diagnosis, specific problem, lab or observation list value, etc. It is also possible to send a BPA to an individual’s EPIC ‘In Basket’, e.g. the study coordinator, who can then contact the patient. You can also set up BPAs to be notified when study participants are administered a drug that would result in their exclusion. BPAs are configured to be displayed only for a specific site, department or user role. When a patient opts not to participate in a study, that particular notification will be suppressed in the future.
The EHR data of patients included in the study can be extracted afterwards (see Data extraction from the EHR for the procedure, conditions and templates). To use these tools, please contact the RDM helpdesk. We will provide advice on using the right application and carry out the cohort identification process together if you wish to use SlicerDicer.
Snowflake is a data warehouse made suitable for the storage, transformation and release of highly confidential (time-series) data originating from Philips bed monitors. Large amounts of vital data obtained via bed monitoring (e.g. continuous blood pressure and ECG data) are now accessible via the Snowflake cloud solution. Good performance and maximum scalability make it possible to query and release these data.
Monitoring data can be used for research, monitoring and dash boarding, e.g. for developing (predictive) algorithms in the context of Health Data Science.
The Adult Intensive Care and Neonatal Intensive Care departments are heavy users of this application and have their own data scientist(s) to extract and analyze the data. Other departments can access the application or request a data extract via the RDM helpdesk.
Data may only be released in compliance with SOP 002 Reuse of health care data for the purpose of research. To make a request, submit the following request form: F01 Health care data extraction request.