EsPRit: ethics committee proposals for Long Term Medical Data Registries in rapidly evolving research fields - a future-proof best practice approach

Given this unique situation, it is exceedingly difficult to draft a MDR-LT ethics proposal in a manner that it guarantees its approval while accounting for the needs and dynamics of such a system. Therefore, an approach is needed which allows necessary and reasonable future changes of such a registry (e.g. of collected data), while still guaranteeing that the MDR-LT adheres to strong ethical standards in medical research and meets all requirements as defined by an ethics committee.

The objective of our work is to present a best practice approach to overcome this dilemma: The EsPRit (EthicS Proposal RegIsTry) approach.

EsPRit defines generic building blocks for a proposal including appropriate methods to parameterize these building blocks and subsequently follows a general guideline compiling ethics proposals for MDR-LT projects. EsPRit can be used in two ways: to create an ethics proposal in the planning phase of the MDR-LT project (without the need to know all the details of the desired registry project in advance); to adapt such a proposal during the lifetime of the MDR-LT whenever substantial aspects (e.g. the clinical questions) are altered by applying the attached methods.

EsPRit is based on a thorough literature review and uses components of the Systematic Planning of Intelligent Reuse of Integrated Clinical Routine Data (SPIRIT) framework for the systematic planning of intelligent reuse of integrated clinical routine data [8]. The concept for EsPRit was designed in a radiotherapeutics research project investigating requirements for a secondary malignancies MDR-LT. EsPRit was refined based on the lessons learned in this case study.

As defined by Arts et al. [9] in 2002, a Medical Data Registry (MDR) is: “a systematic collection of a clearly defined set of health and demographic data for patients with specific health characteristics, held in a central database for a predefined purpose.” Although this definition accounts for the basic purpose of registries it does not reflect recent technological developments such as distributed storage of medical data as, for example, proposed by state of the art electronic health record concepts [10]. A more abstract definition is introduced by Drolet et al., who define five essential characteristics for each MDR [11]:

In addition to these five characteristics Drolet et al. postulated that a kind of the inclusion principle for patients should be taken into considerations for a medical data registry (DZ: chronic disease; TH: acute/interventional).

Accordingly, long-term Medical Data Registries (MDR-LT) have to be understood as a special subset of MDR dedicated to supporting long-term investigations. These MDR-LT have very long follow-up and data-collection periods, typically exceeding 20 years.

Within the field of radiotherapy, major advances have been made throughout the last years in treating patients who suffer from various forms of cancer. Many of these new therapies (e.g. volumetric-modulated arc therapy (VMAT) and intensity-modulated radiation therapy (IMRT)) have one thing in common compared to classical therapies: increased body volume exposure to low dose of radiation. Several studies showed that the development of secondary malignancies correlates with the intensity and distribution of radiation [12‐14]. It is - for now - unclear how these recently developed methods in radiotherapy affect the development of secondary malignancies. Hence, a long-term follow-up of patients is critical in the study of potential effects and their confounders.

SEMPER (Secondary Malignoma - Prospective Evaluation of the radiotherapeutics dose distribution as the cause for induction) [15] was a project part of the COMET Center ONCOTYROL,¹ which was funded by the Austrian Federal Ministries BMVIT/BMWFJ (via FFG) and the Tiroler Zukunftsstiftung/Standortagentur Tirol (SAT). One of the central objectives of SEMPER was to analyze requirements for a MDR-LT, which can be used to study long-term qualitative and quantitative outcomes of these kinds of radiotherapy. Especially long-term effects of an exposure of large body-volumes to low radiation dosage during treatment is an imperative aspect (e.g. increasing incidence of secondary malignancies). In order to reach this goal, a variety of clinical data of participating patients and controls must be collected in a consistent manner. Particularly the long-term perspective of the data collection process must be considered. Secondary malignancies may arise late after initial treatment, several years in the future. Therefore, relevant data must be collected throughout a long period of time (>20 years) as well as complete follow-ups for all included subjects. As patient mobility has increased and given the long observation period of these studies, the data collection process must reflect a multi-center approach that allows cross-site data integration.

The outline and content to be provided by an ethics committee proposal are mainly based on the Declaration of Helsinki and quality standard GCP. However, ethic committees adopt or extend these standards and guidelines to comply with local requirements, specific regulations and the applicable law. Usually, the committees provide templates to support researchers to create ethics proposals.

GCP defines several documents which have to be provided to (independent) ethics committees: “trial protocol(s)/amendment(s), written informed consent form(s) and consent form updates that the investigator proposes for use in the trial, subject recruitment procedures (e.g. advertisements), written information to be provided to subjects, Investigator’s Brochure (IB), available safety information, information about payments and compensation available to subjects, the investigator is current curriculum vitae and/or other documentation evidencing qualifications” [6]. The trial protocol has to contain exact information regarding methods of analysis and data elements to be collected in form of Case Report Forms (CRF).

The recommended structure of a research protocol is outlined by the World Health Organization (WHO) [16]. As with the GCP guidelines, the WHO guidelines require that all data elements, as well as the data analyzing methods, have to be specified prior to the commencing of the study. This neither accounts for a possible change/extension in methods nor for the analysis or for changes in data elements over time.

MDR-LT projects widely use up-to-date software tools [17]. These tools are usually developed following defined software engineering paradigms such as the waterfall model [18], eXtreme Programming (XP) [19], Rational Unified Process (RUP) [20], SCRUM [21] or the Agile Unified Process (AUP) [22]. The base and ongoing input of any of these software engineering paradigms are requirements that have to be provided by the project’s stakeholders or elaborated on together with said stakeholders either at the begin or during implementation. Although the concrete requirements can vary depending on the respective software project, the collection and validation of requirements can be standardized to some extent. The concept of Requirements Engineering (RE) [23, 24] outlines the processes and tasks to manage requirements on a general level (define, document, maintain, validate). The following common sub processes in RE can be identified:

These sub processes are reflected in the EsPRit approach. As depicted before, MDR-LT have to deal with certain uncertainty concerning necessary data elements as well as the relevant research questions and can be subject to various changes during their lifetime. Thus, the base EsPRit model was enriched with components of the SPIRIT framework [8].

SPIRIT is a conceptual best-practice framework for the systematic planning of intelligent reuse of integrated clinical routine data. It can be applied to overcome the gap in situations where reuse of data was not intended. This often leads to the rarity of explicit research questions for secondary use initiatives of clinical data in the planning phase of such data reuse solutions. In many cases, additional and more definite research questions emerge once a critical mass of data has been accumulated and first investigations lead to a crystallization of plausible research hypotheses. SPIRIT suggests an iterative modus operandi to leave scope for development and evolution in such situations, which was adapted for EsPRit.

General MDR-LT requirements collected in a systematic literature review [27] prior to this work were then used to infer and align groups of tasks (building blocks) that have to be carried out to formulate a proper ethics proposal. According to the general requirements, a set of appropriate methods was defined for each building block. The resulting EsPRit model proposes an iterative workflow along the RE sub-processes and defines generic building blocks for an ethics proposal.

To test the feasibility and practicability of EsPRits an ethics committee proposal was compiled for SEMPER. The lessons learned during this case study were then used to further refine the EsPRit approach. A crucial insight that was gained was, for example, that consensus on an initial data set for the MDR-LT was very hard to find. Therefore, the central EsPRit process of electronic case report forms (eCRF) design (see below) was adapted to better support an iterative data set definition starting with a minimal core MDR dataset.

To represent the EsPRit approach (processes, building blocks, methods, results) BPMN2 process modeling is used for specifying business processes. The concept of Swim Lanes was adopted to model the main building blocks; these lanes consist of a pool and a lane. Here, the pool represents the main activity of the block, whereas the lanes summarize tasks to retrieve information. For better readability, the names given to these tasks indicate the information source. Each block produces an outcome (data object). The methods to retrieve and process information of each block are outlined as annotations.

Figure 1 outlines the five building blocks (surrounded by dashed rectangles). Each block contains a main activity (blue rectangles). Green rectangles describe assigned tasks that have to be carried out in order to collect necessary information. For a better readability the task’s names refer to the data sources for this information (S). The methods (M) that can be used to process or collect the information are presented as annotations to the block. The outcomes (O) of the blocks are represented as data objects (blue document icons). For an initial setup in creating an ethics committee proposal the order as outlined in the process model has to be followed. Special attention has to be paid to the block “Context Analysis” (B1) as it derives information of the “Requirements Analysis” (B2) as well as for the “eCRF Design” (B4) and must be carried out in synchronization with these blocks.

The following sections describe each building block (B1 – B5) with its main activity, the information sources, methods and outcome.

At compile time of the ethics committee application, or even after the document for submission had been created, new requirements (e.g. regarding data elements for the MDR-LT) may be detected. Additionally, as outlined before, once completed, the ethics committee application may be subject to change, e.g. due to new data elements that needed to be. To reflect these possibilities, it is proposed that only the affected (e.g. B4 – eCRF Design) and the subsequent blocks (B5 – Overall Concept Creation) must undergo reconsideration, since other blocks are left unaffected. The lifecycle management should be carried out in an iterative manner as proposed by software engineering approaches such as AUP.

If opinions of medical experts differ on fundamental issues like study goals (B1) or necessary data sets to answer the research question (B4) a consensus has to be reached. If a consensus cannot be reached the designers of the MDR-LT have to decide whether to proceed preparing the application or revise the initial goal of the study. Any concerns regarding the feasibility have to be documented in a comprehensible way.

The study designers describe the process on how to e.g. discover MDR-LT requirements to meet the goal of the study. The process should adhere to well proven and documented methods. If an ethics committee decision requires changes to the described process the comments have to be analyzed and if feasible incorporated into the application. Since the process of compiling the application is iterative not all building blocks have to be reconsidered.

At any point in time during the life cycle of a MDR-LT an ethics committee proposal compiled using EsPRit is a fundamental part of the documentation of the registry. If, for example data collection methods change over time the adaption has to be reflected in the ethics committee proposal. Thus, a new decision covering the changes has to be made by the committee. Therefore, at any point in time and for each data item it is well documented how the data was collected and how the data was interpreted.

The EsPRit approach defines building blocks, activities, methods and a process to systematically design MDR-LT in a goal-oriented, standardized and reproducible way. Thus, it can be used to compile the necessary ethics committee proposals for a planned MDR-LT by defining and describing the concrete activities, information sources and methods within each building block. This process-oriented approach follows the idea put forth in medicinal product certification. In contrast to a content-based approach, the outcome of each block must not be defined in advance. A system design adhering to EsPRit will lead to an MDR-LT that will be in accordance to the high ethical standards required by medical research. Thus, a certifying body, e.g. an ethics committee is able to make an informed decision already in the planning phase of the MDR-LT.

However, the decision on whether to permit or decline a research project on the grounds of ethical reasoning is highly dependent on the local implementation of ethical standards. The committee has to support the shift from a content-based approach to one that is process-oriented, ensuring that the research project complies with the implemented ethical standards. Additionally, the commitment to adhere to the defined process outlined by EsPRit is an absolute prerequisite on behalf of the researchers involved in implementing a MDR-LT. Especially a change management process must be implemented in order to reflect the evolution of the MDR-LT during its life cycle. These changes have to be communicated to the ethical committee in an appropriate way. The committee must then revise those changes. In this case, it is not necessary to examine every part of the initial ethical committee application; only changed parts (e.g. added data items in an eCRF) have to be reconsidered. Ethical committees will have to adapt to this new shift by providing, for example, new guidelines that reflect this approach as well as the local requirements.

Compiling an ethics proposal as proposed by EsPRit may add overhead to the initial planning phase of an MDR-LT. Nevertheless, the elaborated building blocks for the ethics proposal can be reused as an initial draft description for the concrete MDR-LT implementation. The inferred overall concept forms the basis of, for example, an implementation request to a software development company building the concrete MDR-LT. The software implementation process is independent of a specific implementation process model as long as the software fulfills the elaborated requirements.

During the initial compilation of the ethics committee proposal in the SEMPER case study, the eCRF was subject of change due to the research question’s vague specification. Additional data elements had to be added during the elaboration of the MDR-LT concept. However, since the Requirements Analysis (B2) and the Requirements Validation (B3) were not affected (as adding new data elements does not affect requirements e.g. regarding data storage, etc.), these blocks had not to be reconsidered. Within the Overall Concept Creation (B5) the outlined agile processes defined by software development allowed for the integration of these new data elements very easily into the final MDR-LT concept. Within SEMPER, utilizing EsPRit as well as the inferred eCRF, an excerpt of the created proposal was used to compile a traditional ethics committee proposal. This decision was made in conjunction with the SEMPER team and the ethics committee in order to ensure on-time implementation of the MDR-LT, since implementing the new approach of compiling ethics committee applications using EsPRit requires further evaluation regarding its feasibility and applicability in medical research routine.

The switch to a process-oriented approach in creating medical committee applications provides a new way of ensuring high ethical standards in medical research. As far as we know, this kind of procedure has not been proposed yet. In conjunction with existing standardizing efforts in the field of data management for clinical trials and MDRs, the complexity of creating an ethics committee proposal using the proposed process-oriented EsPRit approach can be reduced. ECRIN is a standard which provides 115 IT requirements, 107 Data Management requirements and 13 other requirements to provide GCP-compliant data management for clinical trials [1]. Using these requirements as a MDR-LT following the EsPRit approach will reduce the efforts to elaborate the building blocks B2 and B3. Of course, additional requirements reflecting the special setting for which the MDR-LT is to be developed for will have to be considered as well.

A crucial part of a MDR-LT is the integration of existing routine data. Projects mainly focusing on reusing existing routine data for scientific purposes are, for example, “Electronic Health Records for Clinical Research” (EHR4CR)⁶ or “Retrieve from EHR Useful clinical data for Secondary Exploitation” (RE-USE) [38]. Both projects focus, while using standard-based tools and services (e.g. based on Integrating the Healthcare Enterprise (IHE) profiles, etc.), to extract data from electronic health records and provide these data sets to clinical research in a meaningful und useful. Additionally, single-source data entry, collection and cross-system data reuse shall be supported. Especially the RE-USE project focuses on technical solutions to enable semantic integration of different data sets. When using these data sets for clinical research, ethical aspects have to be taken into account as well. Approaches developed within these projects can be integrated within EsPRit to provide technical requirements of the data management in a MDR-LT. The process-oriented approach allows the addition and refinement of the technical methods during the development and runtime of the MDR-LT without the need to rewrite the complete ethics committee proposal.

The added value of EsPRit for MDR-LT projects results in reliable planning and thus, when adhering to this process, an increased compliance to ethical guidelines over time.

Legal aspects such as patient data privacy in long-term studies or the reporting of quality data sets for national health planning tasks must be taken into account in order to use the EsPRit approach in setting up MDR-LTs. It must be reviewed whether the high ethical standards in clinical research are sufficiently reflected. Established ethics committees have to decide whether they accept such proposals or whether additional information must be provided to ensure such things as patient rights. Especially due to changes of the MDR-LT over time, such as in terms of data items to collect, ethic committee proposals created based on the EsPRit approach can reduce the time to provide an adapted application for the committee. Thus far it remains unclear whether those adapted versions are sufficient in ensuring high ethical standards in clinical research. For evaluation of the proposed approach, this long-term perspective must be strongly taken into account.

The initial compilation of an ethics committee proposal using the EsPRit approach may consume additional time and effort in comparison to the traditional manner. Nevertheless, in the long-term it makes sense to spend these additional efforts since such a proposal can support the lifecycle management of MDR-LTs and can be reused for new MDR-LT projects. Only requirements dedicated to the new setting of the MDR-LT being developed would have to be added in such a case. Adapting the building blocks B1 and B4, which are mainly used to reflect the clinical research questions, should be sufficient for most projects.

Nevertheless, within the SEMPER research project, a process-oriented approach was used to create a basic MDR-LT construct, covering the long-term perspective of data collection in order to answer a specific research question, as well as providing a structure to better represent a volatile field in medical research. The result showed that the inferred building blocks and the process-oriented approach of EsPRit provide a good structure using a process-oriented platform in inferring ethical committee applications.