Study Execution

DARWIN EU Studies are executed using a so-called federation approach as shown in the figure below. In this approach the Data Partners in the distributed data network keep full control of their own data. No patient level data is shared with the Coordination Centre (CC) only aggregated study results.

The Study Execution processes are divided into the following Phases: Exploration, Initiation, Implementation, Execution, Dissemination:

Exploration Phase

In the Exploration Phase the EMA is interacting with the Coordination Centre to explore the feasibility of a study request with respect to the preferred design, available data, resources, and timelines.

1. Study Request

Research questions originate from the EU Regulatory Network as a need for information to support the authorisation and monitoring of medicinal products and subsequent decision-making. They may include, for example, queries on the incidence or prevalence of disease in the European population, the drug utilisation patterns for a specific product, the collection of additional data on a potential safety signal, the evaluation of an association between a drug and a clinical outcome or the assessment of the effectiveness of a risk minimisation measure. It is envisaged that in the future, research questions may also originate from other stakeholders such as the European Commission, the European Centre for Disease Prevention and Control (ECDC), Health Technology Assessment (HTA) bodies and payers.

The requester, together with EMA, prepares the Feasibility Assessment Form containing: a short title for the request, the regulatory background motivating the query, any other important information for understanding the query, the product and/or event(s)/outcome(s) of interest, the objective, a proposal for the study design and any relevant design considerations such as the choice of data sources and other methodological aspects. The choice of study design is supported by the Catalogue of Standard Data Analyses providing, for a range of possible analyses with different levels of complexity (Routine repeated, Off-the-shelf, Complex), the implications in terms of databases needed and timelines.

2. Feasibility Assessment

The Coordination Centre then performs a feasibility assessment for EMA review. Its aim is to assess the feasibility of the study based on relevant factors for the concerned study, including, for example, the number of exposed patients, number of events, time period, minimum duration of exposure to be considered, time lag for occurrence of event, feasibility of implementing the study design in the available databases or other criteria depending on the study. The feasibility assessment will in principle only use information already available to CC on the characteristics of databases via pre-calculated Data Dashboards and other relevant information.

If necessary, simple data analyses could be performed in relevant databases available, for example simple count of the number of patients exposed to a medicinal product or who experienced a clinical event over a defined time period. Such data analysis may be performed in a limited number of databases by the Coordination Centre and will not undergo the same quality assurance process as those performed for a full study.

The results of the feasibility analysis are next included in the Feasibility Assessment Form with a recommendation regarding the conduct of the study. The decision to initiate the study is formally made by the EMA.

Initiation Phase

In the Initiation Phase the contractual agreements are setup with the selected Data Partners, the Study Team is formed, and the Conflicts of Interest procedure is applied.

Contractual Agreements

Upon green light to conduct a study, the Coordination Centre is responsible for the contractual agreement and onboarding of Data Partners in the study. A standardised work order is initiated by the DARWIN EU® CC which indicates the agreed timelines and the budget of the study. Each data partner will sign a standard work order agreement which specifies their activities, budget, and timelines. 

Team Formation

The level of involvement of a study team will strongly depend on the Study Type, i.e., simple off the shelf studies may be run near automatically with a fast review step by a small team. For more complex studies a larger team may be proposed by Study Operations. This team consist of data scientists, epidemiologists, and clinical domain experts who are responsible for Evidence Generation and Evidence Synthesis. The team is completed with Data Stewards from the participating Data Partners who will work in a secured collaboration space.

Documentation and assessment of Conflicts of Interest

Potential investigators to be assigned to a DARWIN EU® study shall be requested to declare any existing direct or indirect interests in a pharmaceutical company. These terms are further defined in the EMA Policy 0044. More details on The DARWIN EU Conflict of Interest Procedure can be found in our Principles & Policies section. EMA will review the submitted forms and will approve the team composition.

 

 

Implementation Phase

In the Implementation Phase all the study artefacts are created such as the Study Outline, Study Protocol, Cohort Definitions, Study Package.

Implementation

Study Outline

Once agreement is reached between EMA and the DARWIN EU® CC to conduct a DARWIN EU® study, the Study Outline Form is completed describing aspects such as: Study Team, Data Sources, Choice of study type and related study design including methodological aspects, governance board approval and timelines.

The completed Study Outline Form is next submitted to the EMA for approval. Once the Study Outline Form is approved by the EMA, the protocol for the DARWIN EU® studies is generated.

The Study Outline as approved and signed-off by EMA is archived at the DARWIN EU® CC in a study specific project file.

Study Protocol Development

Once agreement is reached between EMA and the DARWIN EU® CC on the Study Outline, the Study Protocol is generated following the ENCePP code of conduct.

Upon completion of the Draft Study Protocol, this Study Protocol needs to be approved by the EMA. 

Once the Study Protocol is approved and signed-off by EMA, the study protocol is archived at the DARWIN EU® CC in a study specific project file.

Next, the final study protocol is uploaded to the EU PAS Register. 

Kick-off meeting

Following finalisation of the protocol and prior to study execution, a kick-off meeting is organised by the Study Project Manager/Principal Investigator with all study team members to explain the protocol, agree on roles, responsibilities and timelines for study execution. 

Governance Board Approval

Protocol approval at the study sites is often required to comply with the local implementation of relevant laws and directives in the various European states. Each data custodian is responsible for protocol approval from their local governance board prior to study execution as is specified in the Framework Contract with the Data Partner. 

Cohort Construction and Assessment

Where needed, a phenotype for the target and outcome cohorts that work across the included data sources is generated. This is facilitated through the Codelist Generator R package, the Phenotype Library in addition to the Cohort Diagnostics R package. The latter provides an in-depth characterisation of the phenotype definitions, secular trends over time, standard as well as orphan source codes (not included source codes that may be of interest), and cohort participant/s characterisation. 

Study Package Creation

The analysis pipeline for each study type as described within the Catalogue of Standard Data Analysis will be used and only needs to be instantiated with the cohorts of interest and the right parameter settings to obtain the Study Package. For example, for a study to estimate the incidence and prevalence of a health outcome the IncidencePrevalence Package is used and the parameters are set according to the protocol. Before study initiation, test runs of the analytics are performed on a subset of the data sources or on a simulated set of patients and quality control checks are performed. Once all the tests are passed, the final package is released as a private repository in the version- controlled Study Repository. This package is distributed to the data partners for local execution agains their OMOP CDM.

 

Execution Phase

In the Execution Phase the Data Partners execute the Study Package against the OMOP CDM and review the aggregated results. This includes applying quality checks on the study results and a data privacy check to assure that only aggregated data is shared compliant to their local governance procedures. 

Execution Iterations

The data partners locally execute the analytics against the OMOP-CDM and review and approve the aggregated results before returning them to the Coordination Centre. Sometimes multiple execution iterations are performed, and additional fine tuning of the code base is needed. A service desk is available during the study execution for support.

Quality Control

The study results of all data sources are checked after which they are made available to the team in the Digital Research Environment and the Dissemination Phase can start. All results are locked and timestamped for reproducibility and transparency.

Dissemination Phase

In the Dissemination Phase the evidence is synthesised in reports and result dashboards, and potentially in publications. All study artefacts are then retained 

Evidence Synthesis

The results of all data sources are combined to produce the tables and figures needed for the report in the Digital Research Environment. The amount of effort for Evidence Synthesis will strongly depend on the study complexity, i.e., less effort for the “off-the-shelf” and “routine repeated studies” and more effort for the complex studies. The pooling of database-specific results will be pre-specified per protocol, including rules/diagnostics precluding the inclusion of a database and the use of meta-analytic approaches to combine data-specific estimates. The results are reviewed, and conclusions are formed in the context of the prior knowledge if relevant.

Dashboard Instantiation

In most studies executed against the OMOP-CDM, a large amount of study results is generated from many data sources. Therefore, dashboards in the form of a web application will be made to simplify the exploration of the results. All the uploaded study results can be easily combined in a single folder from which the dashboard is instantiated. The web application will either be made publicly available or if necessary, will be made available behind a security layer.

Reporting of Study Results

The study team then writes the report and/or publication. All study reports are written in accordance with the Guideline on good pharmacovigilance (GVP) module VIII, (EMA/813938/2011) and the guidance from ENCePP. The (final) study report(s) are uploaded in the EU PAS Register.

Study findings will preferably be considered for publication as open access. Any publication will be guided by the Uniform Requirements for Manuscripts Submitted to Biomedical Journals: Writing and Editing for Biomedical Publication of the International Committee of Medical Journal Editors (ICMJE). 

Retention

Before closure of the study all generated documents, used analytics, and study results are archived to enable reproducibility.